fast-import.c

   1 #include "builtin.h"
   2 #include "cache.h"
   3 #include "repository.h"
   4 #include "config.h"
   5 #include "lockfile.h"
   6 #include "object.h"
   7 #include "blob.h"
   8 #include "tree.h"
   9 #include "commit.h"
  10 #include "delta.h"
  11 #include "pack.h"
  12 #include "refs.h"
  13 #include "csum-file.h"
  14 #include "quote.h"
  15 #include "dir.h"
  16 #include "run-command.h"
  17 #include "packfile.h"
  18 #include "object-store.h"
  19 #include "mem-pool.h"
  20 #include "commit-reach.h"
  21
  22 #define PACK_ID_BITS 16
  23 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
  24 #define DEPTH_BITS 13
  25 #define MAX_DEPTH ((1<<DEPTH_BITS)-1)
  26
  27 /*
  28  * We abuse the setuid bit on directories to mean "do not delta".
  29  */
  30 #define NO_DELTA S_ISUID
  31
  32 /*
  33  * The amount of additional space required in order to write an object into the
  34  * current pack. This is the hash lengths at the end of the pack, plus the
  35  * length of one object ID.
  36  */
  37 #define PACK_SIZE_THRESHOLD (the_hash_algo->rawsz * 3)
  38
  39 struct object_entry {
  40         struct pack_idx_entry idx;
  41         struct object_entry *next;
  42         uint32_t type : TYPE_BITS,
  43                 pack_id : PACK_ID_BITS,
  44                 depth : DEPTH_BITS;
  45 };
  46
  47 struct object_entry_pool {
  48         struct object_entry_pool *next_pool;
  49         struct object_entry *next_free;
  50         struct object_entry *end;
  51         struct object_entry entries[FLEX_ARRAY]; /* more */
  52 };
  53
  54 struct mark_set {
  55         union {
  56                 struct object_entry *marked[1024];
  57                 struct mark_set *sets[1024];
  58         } data;
  59         unsigned int shift;
  60 };
  61
  62 struct last_object {
  63         struct strbuf data;
  64         off_t offset;
  65         unsigned int depth;
  66         unsigned no_swap : 1;
  67 };
  68
  69 struct atom_str {
  70         struct atom_str *next_atom;
  71         unsigned short str_len;
  72         char str_dat[FLEX_ARRAY]; /* more */
  73 };
  74
  75 struct tree_content;
  76 struct tree_entry {
  77         struct tree_content *tree;
  78         struct atom_str *name;
  79         struct tree_entry_ms {
  80                 uint16_t mode;
  81                 struct object_id oid;
  82         } versions[2];
  83 };
  84
  85 struct tree_content {
  86         unsigned int entry_capacity; /* must match avail_tree_content */
  87         unsigned int entry_count;
  88         unsigned int delta_depth;
  89         struct tree_entry *entries[FLEX_ARRAY]; /* more */
  90 };
  91
  92 struct avail_tree_content {
  93         unsigned int entry_capacity; /* must match tree_content */
  94         struct avail_tree_content *next_avail;
  95 };
  96
  97 struct branch {
  98         struct branch *table_next_branch;
  99         struct branch *active_next_branch;
 100         const char *name;
 101         struct tree_entry branch_tree;
 102         uintmax_t last_commit;
 103         uintmax_t num_notes;
 104         unsigned active : 1;
 105         unsigned delete : 1;
 106         unsigned pack_id : PACK_ID_BITS;
 107         struct object_id oid;
 108 };
 109
 110 struct tag {
 111         struct tag *next_tag;
 112         const char *name;
 113         unsigned int pack_id;
 114         struct object_id oid;
 115 };
 116
 117 struct hash_list {
 118         struct hash_list *next;
 119         struct object_id oid;
 120 };
 121
 122 typedef enum {
 123         WHENSPEC_RAW = 1,
 124         WHENSPEC_RFC2822,
 125         WHENSPEC_NOW
 126 } whenspec_type;
 127
 128 struct recent_command {
 129         struct recent_command *prev;
 130         struct recent_command *next;
 131         char *buf;
 132 };
 133
 134 /* Configured limits on output */
 135 static unsigned long max_depth = 50;
 136 static off_t max_packsize;
 137 static int unpack_limit = 100;
 138 static int force_update;
 139
 140 /* Stats and misc. counters */
 141 static uintmax_t alloc_count;
 142 static uintmax_t marks_set_count;
 143 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 144 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 145 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 146 static uintmax_t delta_count_attempts_by_type[1 << TYPE_BITS];
 147 static unsigned long object_count;
 148 static unsigned long branch_count;
 149 static unsigned long branch_load_count;
 150 static int failure;
 151 static FILE *pack_edges;
 152 static unsigned int show_stats = 1;
 153 static int global_argc;
 154 static const char **global_argv;
 155
 156 /* Memory pools */
 157 static struct mem_pool fi_mem_pool =  {NULL, 2*1024*1024 -
 158                                        sizeof(struct mp_block), 0 };
 159
 160 /* Atom management */
 161 static unsigned int atom_table_sz = 4451;
 162 static unsigned int atom_cnt;
 163 static struct atom_str **atom_table;
 164
 165 /* The .pack file being generated */
 166 static struct pack_idx_option pack_idx_opts;
 167 static unsigned int pack_id;
 168 static struct hashfile *pack_file;
 169 static struct packed_git *pack_data;
 170 static struct packed_git **all_packs;
 171 static off_t pack_size;
 172
 173 /* Table of objects we've written. */
 174 static unsigned int object_entry_alloc = 5000;
 175 static struct object_entry_pool *blocks;
 176 static struct object_entry *object_table[1 << 16];
 177 static struct mark_set *marks;
 178 static const char *export_marks_file;
 179 static const char *import_marks_file;
 180 static int import_marks_file_from_stream;
 181 static int import_marks_file_ignore_missing;
 182 static int import_marks_file_done;
 183 static int relative_marks_paths;
 184
 185 /* Our last blob */
 186 static struct last_object last_blob = { STRBUF_INIT, 0, 0, 0 };
 187
 188 /* Tree management */
 189 static unsigned int tree_entry_alloc = 1000;
 190 static void *avail_tree_entry;
 191 static unsigned int avail_tree_table_sz = 100;
 192 static struct avail_tree_content **avail_tree_table;
 193 static size_t tree_entry_allocd;
 194 static struct strbuf old_tree = STRBUF_INIT;
 195 static struct strbuf new_tree = STRBUF_INIT;
 196
 197 /* Branch data */
 198 static unsigned long max_active_branches = 5;
 199 static unsigned long cur_active_branches;
 200 static unsigned long branch_table_sz = 1039;
 201 static struct branch **branch_table;
 202 static struct branch *active_branches;
 203
 204 /* Tag data */
 205 static struct tag *first_tag;
 206 static struct tag *last_tag;
 207
 208 /* Input stream parsing */
 209 static whenspec_type whenspec = WHENSPEC_RAW;
 210 static struct strbuf command_buf = STRBUF_INIT;
 211 static int unread_command_buf;
 212 static struct recent_command cmd_hist = {&cmd_hist, &cmd_hist, NULL};
 213 static struct recent_command *cmd_tail = &cmd_hist;
 214 static struct recent_command *rc_free;
 215 static unsigned int cmd_save = 100;
 216 static uintmax_t next_mark;
 217 static struct strbuf new_data = STRBUF_INIT;
 218 static int seen_data_command;
 219 static int require_explicit_termination;
 220 static int allow_unsafe_features;
 221
 222 /* Signal handling */
 223 static volatile sig_atomic_t checkpoint_requested;
 224
 225 /* Where to write output of cat-blob commands */
 226 static int cat_blob_fd = STDOUT_FILENO;
 227
 228 static void parse_argv(void);
 229 static void parse_get_mark(const char *p);
 230 static void parse_cat_blob(const char *p);
 231 static void parse_ls(const char *p, struct branch *b);
 232
 233 static void write_branch_report(FILE *rpt, struct branch *b)
 234 {
 235         fprintf(rpt, "%s:\n", b->name);
 236
 237         fprintf(rpt, "  status      :");
 238         if (b->active)
 239                 fputs(" active", rpt);
 240         if (b->branch_tree.tree)
 241                 fputs(" loaded", rpt);
 242         if (is_null_oid(&b->branch_tree.versions[1].oid))
 243                 fputs(" dirty", rpt);
 244         fputc('\n', rpt);
 245
 246         fprintf(rpt, "  tip commit  : %s\n", oid_to_hex(&b->oid));
 247         fprintf(rpt, "  old tree    : %s\n",
 248                 oid_to_hex(&b->branch_tree.versions[0].oid));
 249         fprintf(rpt, "  cur tree    : %s\n",
 250                 oid_to_hex(&b->branch_tree.versions[1].oid));
 251         fprintf(rpt, "  commit clock: %" PRIuMAX "\n", b->last_commit);
 252
 253         fputs("  last pack   : ", rpt);
 254         if (b->pack_id < MAX_PACK_ID)
 255                 fprintf(rpt, "%u", b->pack_id);
 256         fputc('\n', rpt);
 257
 258         fputc('\n', rpt);
 259 }
 260
 261 static void dump_marks_helper(FILE *, uintmax_t, struct mark_set *);
 262
 263 static void write_crash_report(const char *err)
 264 {
 265         char *loc = git_pathdup("fast_import_crash_%"PRIuMAX, (uintmax_t) getpid());
 266         FILE *rpt = fopen(loc, "w");
 267         struct branch *b;
 268         unsigned long lu;
 269         struct recent_command *rc;
 270
 271         if (!rpt) {
 272                 error_errno("can't write crash report %s", loc);
 273                 free(loc);
 274                 return;
 275         }
 276
 277         fprintf(stderr, "fast-import: dumping crash report to %s\n", loc);
 278
 279         fprintf(rpt, "fast-import crash report:\n");
 280         fprintf(rpt, "    fast-import process: %"PRIuMAX"\n", (uintmax_t) getpid());
 281         fprintf(rpt, "    parent process     : %"PRIuMAX"\n", (uintmax_t) getppid());
 282         fprintf(rpt, "    at %s\n", show_date(time(NULL), 0, DATE_MODE(ISO8601)));
 283         fputc('\n', rpt);
 284
 285         fputs("fatal: ", rpt);
 286         fputs(err, rpt);
 287         fputc('\n', rpt);
 288
 289         fputc('\n', rpt);
 290         fputs("Most Recent Commands Before Crash\n", rpt);
 291         fputs("---------------------------------\n", rpt);
 292         for (rc = cmd_hist.next; rc != &cmd_hist; rc = rc->next) {
 293                 if (rc->next == &cmd_hist)
 294                         fputs("* ", rpt);
 295                 else
 296                         fputs("  ", rpt);
 297                 fputs(rc->buf, rpt);
 298                 fputc('\n', rpt);
 299         }
 300
 301         fputc('\n', rpt);
 302         fputs("Active Branch LRU\n", rpt);
 303         fputs("-----------------\n", rpt);
 304         fprintf(rpt, "    active_branches = %lu cur, %lu max\n",
 305                 cur_active_branches,
 306                 max_active_branches);
 307         fputc('\n', rpt);
 308         fputs("  pos  clock name\n", rpt);
 309         fputs("  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n", rpt);
 310         for (b = active_branches, lu = 0; b; b = b->active_next_branch)
 311                 fprintf(rpt, "  %2lu) %6" PRIuMAX" %s\n",
 312                         ++lu, b->last_commit, b->name);
 313
 314         fputc('\n', rpt);
 315         fputs("Inactive Branches\n", rpt);
 316         fputs("-----------------\n", rpt);
 317         for (lu = 0; lu < branch_table_sz; lu++) {
 318                 for (b = branch_table[lu]; b; b = b->table_next_branch)
 319                         write_branch_report(rpt, b);
 320         }
 321
 322         if (first_tag) {
 323                 struct tag *tg;
 324                 fputc('\n', rpt);
 325                 fputs("Annotated Tags\n", rpt);
 326                 fputs("--------------\n", rpt);
 327                 for (tg = first_tag; tg; tg = tg->next_tag) {
 328                         fputs(oid_to_hex(&tg->oid), rpt);
 329                         fputc(' ', rpt);
 330                         fputs(tg->name, rpt);
 331                         fputc('\n', rpt);
 332                 }
 333         }
 334
 335         fputc('\n', rpt);
 336         fputs("Marks\n", rpt);
 337         fputs("-----\n", rpt);
 338         if (export_marks_file)
 339                 fprintf(rpt, "  exported to %s\n", export_marks_file);
 340         else
 341                 dump_marks_helper(rpt, 0, marks);
 342
 343         fputc('\n', rpt);
 344         fputs("-------------------\n", rpt);
 345         fputs("END OF CRASH REPORT\n", rpt);
 346         fclose(rpt);
 347         free(loc);
 348 }
 349
 350 static void end_packfile(void);
 351 static void unkeep_all_packs(void);
 352 static void dump_marks(void);
 353
 354 static NORETURN void die_nicely(const char *err, va_list params)
 355 {
 356         static int zombie;
 357         char message[2 * PATH_MAX];
 358
 359         vsnprintf(message, sizeof(message), err, params);
 360         fputs("fatal: ", stderr);
 361         fputs(message, stderr);
 362         fputc('\n', stderr);
 363
 364         if (!zombie) {
 365                 zombie = 1;
 366                 write_crash_report(message);
 367                 end_packfile();
 368                 unkeep_all_packs();
 369                 dump_marks();
 370         }
 371         exit(128);
 372 }
 373
 374 #ifndef SIGUSR1 /* Windows, for example */
 375
 376 static void set_checkpoint_signal(void)
 377 {
 378 }
 379
 380 #else
 381
 382 static void checkpoint_signal(int signo)
 383 {
 384         checkpoint_requested = 1;
 385 }
 386
 387 static void set_checkpoint_signal(void)
 388 {
 389         struct sigaction sa;
 390
 391         memset(&sa, 0, sizeof(sa));
 392         sa.sa_handler = checkpoint_signal;
 393         sigemptyset(&sa.sa_mask);
 394         sa.sa_flags = SA_RESTART;
 395         sigaction(SIGUSR1, &sa, NULL);
 396 }
 397
 398 #endif
 399
 400 static void alloc_objects(unsigned int cnt)
 401 {
 402         struct object_entry_pool *b;
 403
 404         b = xmalloc(sizeof(struct object_entry_pool)
 405                 + cnt * sizeof(struct object_entry));
 406         b->next_pool = blocks;
 407         b->next_free = b->entries;
 408         b->end = b->entries + cnt;
 409         blocks = b;
 410         alloc_count += cnt;
 411 }
 412
 413 static struct object_entry *new_object(struct object_id *oid)
 414 {
 415         struct object_entry *e;
 416
 417         if (blocks->next_free == blocks->end)
 418                 alloc_objects(object_entry_alloc);
 419
 420         e = blocks->next_free++;
 421         oidcpy(&e->idx.oid, oid);
 422         return e;
 423 }
 424
 425 static struct object_entry *find_object(struct object_id *oid)
 426 {
 427         unsigned int h = oid->hash[0] << 8 | oid->hash[1];
 428         struct object_entry *e;
 429         for (e = object_table[h]; e; e = e->next)
 430                 if (oideq(oid, &e->idx.oid))
 431                         return e;
 432         return NULL;
 433 }
 434
 435 static struct object_entry *insert_object(struct object_id *oid)
 436 {
 437         unsigned int h = oid->hash[0] << 8 | oid->hash[1];
 438         struct object_entry *e = object_table[h];
 439
 440         while (e) {
 441                 if (oideq(oid, &e->idx.oid))
 442                         return e;
 443                 e = e->next;
 444         }
 445
 446         e = new_object(oid);
 447         e->next = object_table[h];
 448         e->idx.offset = 0;
 449         object_table[h] = e;
 450         return e;
 451 }
 452
 453 static void invalidate_pack_id(unsigned int id)
 454 {
 455         unsigned int h;
 456         unsigned long lu;
 457         struct tag *t;
 458
 459         for (h = 0; h < ARRAY_SIZE(object_table); h++) {
 460                 struct object_entry *e;
 461
 462                 for (e = object_table[h]; e; e = e->next)
 463                         if (e->pack_id == id)
 464                                 e->pack_id = MAX_PACK_ID;
 465         }
 466
 467         for (lu = 0; lu < branch_table_sz; lu++) {
 468                 struct branch *b;
 469
 470                 for (b = branch_table[lu]; b; b = b->table_next_branch)
 471                         if (b->pack_id == id)
 472                                 b->pack_id = MAX_PACK_ID;
 473         }
 474
 475         for (t = first_tag; t; t = t->next_tag)
 476                 if (t->pack_id == id)
 477                         t->pack_id = MAX_PACK_ID;
 478 }
 479
 480 static unsigned int hc_str(const char *s, size_t len)
 481 {
 482         unsigned int r = 0;
 483         while (len-- > 0)
 484                 r = r * 31 + *s++;
 485         return r;
 486 }
 487
 488 static char *pool_strdup(const char *s)
 489 {
 490         size_t len = strlen(s) + 1;
 491         char *r = mem_pool_alloc(&fi_mem_pool, len);
 492         memcpy(r, s, len);
 493         return r;
 494 }
 495
 496 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 497 {
 498         struct mark_set *s = marks;
 499         while ((idnum >> s->shift) >= 1024) {
 500                 s = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
 501                 s->shift = marks->shift + 10;
 502                 s->data.sets[0] = marks;
 503                 marks = s;
 504         }
 505         while (s->shift) {
 506                 uintmax_t i = idnum >> s->shift;
 507                 idnum -= i << s->shift;
 508                 if (!s->data.sets[i]) {
 509                         s->data.sets[i] = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
 510                         s->data.sets[i]->shift = s->shift - 10;
 511                 }
 512                 s = s->data.sets[i];
 513         }
 514         if (!s->data.marked[idnum])
 515                 marks_set_count++;
 516         s->data.marked[idnum] = oe;
 517 }
 518
 519 static struct object_entry *find_mark(uintmax_t idnum)
 520 {
 521         uintmax_t orig_idnum = idnum;
 522         struct mark_set *s = marks;
 523         struct object_entry *oe = NULL;
 524         if ((idnum >> s->shift) < 1024) {
 525                 while (s && s->shift) {
 526                         uintmax_t i = idnum >> s->shift;
 527                         idnum -= i << s->shift;
 528                         s = s->data.sets[i];
 529                 }
 530                 if (s)
 531                         oe = s->data.marked[idnum];
 532         }
 533         if (!oe)
 534                 die("mark :%" PRIuMAX " not declared", orig_idnum);
 535         return oe;
 536 }
 537
 538 static struct atom_str *to_atom(const char *s, unsigned short len)
 539 {
 540         unsigned int hc = hc_str(s, len) % atom_table_sz;
 541         struct atom_str *c;
 542
 543         for (c = atom_table[hc]; c; c = c->next_atom)
 544                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 545                         return c;
 546
 547         c = mem_pool_alloc(&fi_mem_pool, sizeof(struct atom_str) + len + 1);
 548         c->str_len = len;
 549         memcpy(c->str_dat, s, len);
 550         c->str_dat[len] = 0;
 551         c->next_atom = atom_table[hc];
 552         atom_table[hc] = c;
 553         atom_cnt++;
 554         return c;
 555 }
 556
 557 static struct branch *lookup_branch(const char *name)
 558 {
 559         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 560         struct branch *b;
 561
 562         for (b = branch_table[hc]; b; b = b->table_next_branch)
 563                 if (!strcmp(name, b->name))
 564                         return b;
 565         return NULL;
 566 }
 567
 568 static struct branch *new_branch(const char *name)
 569 {
 570         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 571         struct branch *b = lookup_branch(name);
 572
 573         if (b)
 574                 die("Invalid attempt to create duplicate branch: %s", name);
 575         if (check_refname_format(name, REFNAME_ALLOW_ONELEVEL))
 576                 die("Branch name doesn't conform to GIT standards: %s", name);
 577
 578         b = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct branch));
 579         b->name = pool_strdup(name);
 580         b->table_next_branch = branch_table[hc];
 581         b->branch_tree.versions[0].mode = S_IFDIR;
 582         b->branch_tree.versions[1].mode = S_IFDIR;
 583         b->num_notes = 0;
 584         b->active = 0;
 585         b->pack_id = MAX_PACK_ID;
 586         branch_table[hc] = b;
 587         branch_count++;
 588         return b;
 589 }
 590
 591 static unsigned int hc_entries(unsigned int cnt)
 592 {
 593         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 594         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 595 }
 596
 597 static struct tree_content *new_tree_content(unsigned int cnt)
 598 {
 599         struct avail_tree_content *f, *l = NULL;
 600         struct tree_content *t;
 601         unsigned int hc = hc_entries(cnt);
 602
 603         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 604                 if (f->entry_capacity >= cnt)
 605                         break;
 606
 607         if (f) {
 608                 if (l)
 609                         l->next_avail = f->next_avail;
 610                 else
 611                         avail_tree_table[hc] = f->next_avail;
 612         } else {
 613                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 614                 f = mem_pool_alloc(&fi_mem_pool, sizeof(*t) + sizeof(t->entries[0]) * cnt);
 615                 f->entry_capacity = cnt;
 616         }
 617
 618         t = (struct tree_content*)f;
 619         t->entry_count = 0;
 620         t->delta_depth = 0;
 621         return t;
 622 }
 623
 624 static void release_tree_entry(struct tree_entry *e);
 625 static void release_tree_content(struct tree_content *t)
 626 {
 627         struct avail_tree_content *f = (struct avail_tree_content*)t;
 628         unsigned int hc = hc_entries(f->entry_capacity);
 629         f->next_avail = avail_tree_table[hc];
 630         avail_tree_table[hc] = f;
 631 }
 632
 633 static void release_tree_content_recursive(struct tree_content *t)
 634 {
 635         unsigned int i;
 636         for (i = 0; i < t->entry_count; i++)
 637                 release_tree_entry(t->entries[i]);
 638         release_tree_content(t);
 639 }
 640
 641 static struct tree_content *grow_tree_content(
 642         struct tree_content *t,
 643         int amt)
 644 {
 645         struct tree_content *r = new_tree_content(t->entry_count + amt);
 646         r->entry_count = t->entry_count;
 647         r->delta_depth = t->delta_depth;
 648         COPY_ARRAY(r->entries, t->entries, t->entry_count);
 649         release_tree_content(t);
 650         return r;
 651 }
 652
 653 static struct tree_entry *new_tree_entry(void)
 654 {
 655         struct tree_entry *e;
 656
 657         if (!avail_tree_entry) {
 658                 unsigned int n = tree_entry_alloc;
 659                 tree_entry_allocd += n * sizeof(struct tree_entry);
 660                 ALLOC_ARRAY(e, n);
 661                 avail_tree_entry = e;
 662                 while (n-- > 1) {
 663                         *((void**)e) = e + 1;
 664                         e++;
 665                 }
 666                 *((void**)e) = NULL;
 667         }
 668
 669         e = avail_tree_entry;
 670         avail_tree_entry = *((void**)e);
 671         return e;
 672 }
 673
 674 static void release_tree_entry(struct tree_entry *e)
 675 {
 676         if (e->tree)
 677                 release_tree_content_recursive(e->tree);
 678         *((void**)e) = avail_tree_entry;
 679         avail_tree_entry = e;
 680 }
 681
 682 static struct tree_content *dup_tree_content(struct tree_content *s)
 683 {
 684         struct tree_content *d;
 685         struct tree_entry *a, *b;
 686         unsigned int i;
 687
 688         if (!s)
 689                 return NULL;
 690         d = new_tree_content(s->entry_count);
 691         for (i = 0; i < s->entry_count; i++) {
 692                 a = s->entries[i];
 693                 b = new_tree_entry();
 694                 memcpy(b, a, sizeof(*a));
 695                 if (a->tree && is_null_oid(&b->versions[1].oid))
 696                         b->tree = dup_tree_content(a->tree);
 697                 else
 698                         b->tree = NULL;
 699                 d->entries[i] = b;
 700         }
 701         d->entry_count = s->entry_count;
 702         d->delta_depth = s->delta_depth;
 703
 704         return d;
 705 }
 706
 707 static void start_packfile(void)
 708 {
 709         struct strbuf tmp_file = STRBUF_INIT;
 710         struct packed_git *p;
 711         struct pack_header hdr;
 712         int pack_fd;
 713
 714         pack_fd = odb_mkstemp(&tmp_file, "pack/tmp_pack_XXXXXX");
 715         FLEX_ALLOC_STR(p, pack_name, tmp_file.buf);
 716         strbuf_release(&tmp_file);
 717
 718         p->pack_fd = pack_fd;
 719         p->do_not_close = 1;
 720         pack_file = hashfd(pack_fd, p->pack_name);
 721
 722         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 723         hdr.hdr_version = htonl(2);
 724         hdr.hdr_entries = 0;
 725         hashwrite(pack_file, &hdr, sizeof(hdr));
 726
 727         pack_data = p;
 728         pack_size = sizeof(hdr);
 729         object_count = 0;
 730
 731         REALLOC_ARRAY(all_packs, pack_id + 1);
 732         all_packs[pack_id] = p;
 733 }
 734
 735 static const char *create_index(void)
 736 {
 737         const char *tmpfile;
 738         struct pack_idx_entry **idx, **c, **last;
 739         struct object_entry *e;
 740         struct object_entry_pool *o;
 741
 742         /* Build the table of object IDs. */
 743         ALLOC_ARRAY(idx, object_count);
 744         c = idx;
 745         for (o = blocks; o; o = o->next_pool)
 746                 for (e = o->next_free; e-- != o->entries;)
 747                         if (pack_id == e->pack_id)
 748                                 *c++ = &e->idx;
 749         last = idx + object_count;
 750         if (c != last)
 751                 die("internal consistency error creating the index");
 752
 753         tmpfile = write_idx_file(NULL, idx, object_count, &pack_idx_opts,
 754                                  pack_data->hash);
 755         free(idx);
 756         return tmpfile;
 757 }
 758
 759 static char *keep_pack(const char *curr_index_name)
 760 {
 761         static const char *keep_msg = "fast-import";
 762         struct strbuf name = STRBUF_INIT;
 763         int keep_fd;
 764
 765         odb_pack_name(&name, pack_data->hash, "keep");
 766         keep_fd = odb_pack_keep(name.buf);
 767         if (keep_fd < 0)
 768                 die_errno("cannot create keep file");
 769         write_or_die(keep_fd, keep_msg, strlen(keep_msg));
 770         if (close(keep_fd))
 771                 die_errno("failed to write keep file");
 772
 773         odb_pack_name(&name, pack_data->hash, "pack");
 774         if (finalize_object_file(pack_data->pack_name, name.buf))
 775                 die("cannot store pack file");
 776
 777         odb_pack_name(&name, pack_data->hash, "idx");
 778         if (finalize_object_file(curr_index_name, name.buf))
 779                 die("cannot store index file");
 780         free((void *)curr_index_name);
 781         return strbuf_detach(&name, NULL);
 782 }
 783
 784 static void unkeep_all_packs(void)
 785 {
 786         struct strbuf name = STRBUF_INIT;
 787         int k;
 788
 789         for (k = 0; k < pack_id; k++) {
 790                 struct packed_git *p = all_packs[k];
 791                 odb_pack_name(&name, p->hash, "keep");
 792                 unlink_or_warn(name.buf);
 793         }
 794         strbuf_release(&name);
 795 }
 796
 797 static int loosen_small_pack(const struct packed_git *p)
 798 {
 799         struct child_process unpack = CHILD_PROCESS_INIT;
 800
 801         if (lseek(p->pack_fd, 0, SEEK_SET) < 0)
 802                 die_errno("Failed seeking to start of '%s'", p->pack_name);
 803
 804         unpack.in = p->pack_fd;
 805         unpack.git_cmd = 1;
 806         unpack.stdout_to_stderr = 1;
 807         argv_array_push(&unpack.args, "unpack-objects");
 808         if (!show_stats)
 809                 argv_array_push(&unpack.args, "-q");
 810
 811         return run_command(&unpack);
 812 }
 813
 814 static void end_packfile(void)
 815 {
 816         static int running;
 817
 818         if (running || !pack_data)
 819                 return;
 820
 821         running = 1;
 822         clear_delta_base_cache();
 823         if (object_count) {
 824                 struct packed_git *new_p;
 825                 struct object_id cur_pack_oid;
 826                 char *idx_name;
 827                 int i;
 828                 struct branch *b;
 829                 struct tag *t;
 830
 831                 close_pack_windows(pack_data);
 832                 finalize_hashfile(pack_file, cur_pack_oid.hash, 0);
 833                 fixup_pack_header_footer(pack_data->pack_fd, pack_data->hash,
 834                                          pack_data->pack_name, object_count,
 835                                          cur_pack_oid.hash, pack_size);
 836
 837                 if (object_count <= unpack_limit) {
 838                         if (!loosen_small_pack(pack_data)) {
 839                                 invalidate_pack_id(pack_id);
 840                                 goto discard_pack;
 841                         }
 842                 }
 843
 844                 close(pack_data->pack_fd);
 845                 idx_name = keep_pack(create_index());
 846
 847                 /* Register the packfile with core git's machinery. */
 848                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 849                 if (!new_p)
 850                         die("core git rejected index %s", idx_name);
 851                 all_packs[pack_id] = new_p;
 852                 install_packed_git(the_repository, new_p);
 853                 free(idx_name);
 854
 855                 /* Print the boundary */
 856                 if (pack_edges) {
 857                         fprintf(pack_edges, "%s:", new_p->pack_name);
 858                         for (i = 0; i < branch_table_sz; i++) {
 859                                 for (b = branch_table[i]; b; b = b->table_next_branch) {
 860                                         if (b->pack_id == pack_id)
 861                                                 fprintf(pack_edges, " %s",
 862                                                         oid_to_hex(&b->oid));
 863                                 }
 864                         }
 865                         for (t = first_tag; t; t = t->next_tag) {
 866                                 if (t->pack_id == pack_id)
 867                                         fprintf(pack_edges, " %s",
 868                                                 oid_to_hex(&t->oid));
 869                         }
 870                         fputc('\n', pack_edges);
 871                         fflush(pack_edges);
 872                 }
 873
 874                 pack_id++;
 875         }
 876         else {
 877 discard_pack:
 878                 close(pack_data->pack_fd);
 879                 unlink_or_warn(pack_data->pack_name);
 880         }
 881         FREE_AND_NULL(pack_data);
 882         running = 0;
 883
 884         /* We can't carry a delta across packfiles. */
 885         strbuf_release(&last_blob.data);
 886         last_blob.offset = 0;
 887         last_blob.depth = 0;
 888 }
 889
 890 static void cycle_packfile(void)
 891 {
 892         end_packfile();
 893         start_packfile();
 894 }
 895
 896 static int store_object(
 897         enum object_type type,
 898         struct strbuf *dat,
 899         struct last_object *last,
 900         struct object_id *oidout,
 901         uintmax_t mark)
 902 {
 903         void *out, *delta;
 904         struct object_entry *e;
 905         unsigned char hdr[96];
 906         struct object_id oid;
 907         unsigned long hdrlen, deltalen;
 908         git_hash_ctx c;
 909         git_zstream s;
 910
 911         hdrlen = xsnprintf((char *)hdr, sizeof(hdr), "%s %lu",
 912                            type_name(type), (unsigned long)dat->len) + 1;
 913         the_hash_algo->init_fn(&c);
 914         the_hash_algo->update_fn(&c, hdr, hdrlen);
 915         the_hash_algo->update_fn(&c, dat->buf, dat->len);
 916         the_hash_algo->final_fn(oid.hash, &c);
 917         if (oidout)
 918                 oidcpy(oidout, &oid);
 919
 920         e = insert_object(&oid);
 921         if (mark)
 922                 insert_mark(mark, e);
 923         if (e->idx.offset) {
 924                 duplicate_count_by_type[type]++;
 925                 return 1;
 926         } else if (find_sha1_pack(oid.hash,
 927                                   get_all_packs(the_repository))) {
 928                 e->type = type;
 929                 e->pack_id = MAX_PACK_ID;
 930                 e->idx.offset = 1; /* just not zero! */
 931                 duplicate_count_by_type[type]++;
 932                 return 1;
 933         }
 934
 935         if (last && last->data.len && last->data.buf && last->depth < max_depth
 936                 && dat->len > the_hash_algo->rawsz) {
 937
 938                 delta_count_attempts_by_type[type]++;
 939                 delta = diff_delta(last->data.buf, last->data.len,
 940                         dat->buf, dat->len,
 941                         &deltalen, dat->len - the_hash_algo->rawsz);
 942         } else
 943                 delta = NULL;
 944
 945         git_deflate_init(&s, pack_compression_level);
 946         if (delta) {
 947                 s.next_in = delta;
 948                 s.avail_in = deltalen;
 949         } else {
 950                 s.next_in = (void *)dat->buf;
 951                 s.avail_in = dat->len;
 952         }
 953         s.avail_out = git_deflate_bound(&s, s.avail_in);
 954         s.next_out = out = xmalloc(s.avail_out);
 955         while (git_deflate(&s, Z_FINISH) == Z_OK)
 956                 ; /* nothing */
 957         git_deflate_end(&s);
 958
 959         /* Determine if we should auto-checkpoint. */
 960         if ((max_packsize
 961                 && (pack_size + PACK_SIZE_THRESHOLD + s.total_out) > max_packsize)
 962                 || (pack_size + PACK_SIZE_THRESHOLD + s.total_out) < pack_size) {
 963
 964                 /* This new object needs to *not* have the current pack_id. */
 965                 e->pack_id = pack_id + 1;
 966                 cycle_packfile();
 967
 968                 /* We cannot carry a delta into the new pack. */
 969                 if (delta) {
 970                         FREE_AND_NULL(delta);
 971
 972                         git_deflate_init(&s, pack_compression_level);
 973                         s.next_in = (void *)dat->buf;
 974                         s.avail_in = dat->len;
 975                         s.avail_out = git_deflate_bound(&s, s.avail_in);
 976                         s.next_out = out = xrealloc(out, s.avail_out);
 977                         while (git_deflate(&s, Z_FINISH) == Z_OK)
 978                                 ; /* nothing */
 979                         git_deflate_end(&s);
 980                 }
 981         }
 982
 983         e->type = type;
 984         e->pack_id = pack_id;
 985         e->idx.offset = pack_size;
 986         object_count++;
 987         object_count_by_type[type]++;
 988
 989         crc32_begin(pack_file);
 990
 991         if (delta) {
 992                 off_t ofs = e->idx.offset - last->offset;
 993                 unsigned pos = sizeof(hdr) - 1;
 994
 995                 delta_count_by_type[type]++;
 996                 e->depth = last->depth + 1;
 997
 998                 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
 999                                                       OBJ_OFS_DELTA, deltalen);
1000                 hashwrite(pack_file, hdr, hdrlen);
1001                 pack_size += hdrlen;
1002
1003                 hdr[pos] = ofs & 127;
1004                 while (ofs >>= 7)
1005                         hdr[--pos] = 128 | (--ofs & 127);
1006                 hashwrite(pack_file, hdr + pos, sizeof(hdr) - pos);
1007                 pack_size += sizeof(hdr) - pos;
1008         } else {
1009                 e->depth = 0;
1010                 hdrlen = encode_in_pack_object_header(hdr, sizeof(hdr),
1011                                                       type, dat->len);
1012                 hashwrite(pack_file, hdr, hdrlen);
1013                 pack_size += hdrlen;
1014         }
1015
1016         hashwrite(pack_file, out, s.total_out);
1017         pack_size += s.total_out;
1018
1019         e->idx.crc32 = crc32_end(pack_file);
1020
1021         free(out);
1022         free(delta);
1023         if (last) {
1024                 if (last->no_swap) {
1025                         last->data = *dat;
1026                 } else {
1027                         strbuf_swap(&last->data, dat);
1028                 }
1029                 last->offset = e->idx.offset;
1030                 last->depth = e->depth;
1031         }
1032         return 0;
1033 }
1034
1035 static void truncate_pack(struct hashfile_checkpoint *checkpoint)
1036 {
1037         if (hashfile_truncate(pack_file, checkpoint))
1038                 die_errno("cannot truncate pack to skip duplicate");
1039         pack_size = checkpoint->offset;
1040 }
1041
1042 static void stream_blob(uintmax_t len, struct object_id *oidout, uintmax_t mark)
1043 {
1044         size_t in_sz = 64 * 1024, out_sz = 64 * 1024;
1045         unsigned char *in_buf = xmalloc(in_sz);
1046         unsigned char *out_buf = xmalloc(out_sz);
1047         struct object_entry *e;
1048         struct object_id oid;
1049         unsigned long hdrlen;
1050         off_t offset;
1051         git_hash_ctx c;
1052         git_zstream s;
1053         struct hashfile_checkpoint checkpoint;
1054         int status = Z_OK;
1055
1056         /* Determine if we should auto-checkpoint. */
1057         if ((max_packsize
1058                 && (pack_size + PACK_SIZE_THRESHOLD + len) > max_packsize)
1059                 || (pack_size + PACK_SIZE_THRESHOLD + len) < pack_size)
1060                 cycle_packfile();
1061
1062         hashfile_checkpoint(pack_file, &checkpoint);
1063         offset = checkpoint.offset;
1064
1065         hdrlen = xsnprintf((char *)out_buf, out_sz, "blob %" PRIuMAX, len) + 1;
1066
1067         the_hash_algo->init_fn(&c);
1068         the_hash_algo->update_fn(&c, out_buf, hdrlen);
1069
1070         crc32_begin(pack_file);
1071
1072         git_deflate_init(&s, pack_compression_level);
1073
1074         hdrlen = encode_in_pack_object_header(out_buf, out_sz, OBJ_BLOB, len);
1075
1076         s.next_out = out_buf + hdrlen;
1077         s.avail_out = out_sz - hdrlen;
1078
1079         while (status != Z_STREAM_END) {
1080                 if (0 < len && !s.avail_in) {
1081                         size_t cnt = in_sz < len ? in_sz : (size_t)len;
1082                         size_t n = fread(in_buf, 1, cnt, stdin);
1083                         if (!n && feof(stdin))
1084                                 die("EOF in data (%" PRIuMAX " bytes remaining)", len);
1085
1086                         the_hash_algo->update_fn(&c, in_buf, n);
1087                         s.next_in = in_buf;
1088                         s.avail_in = n;
1089                         len -= n;
1090                 }
1091
1092                 status = git_deflate(&s, len ? 0 : Z_FINISH);
1093
1094                 if (!s.avail_out || status == Z_STREAM_END) {
1095                         size_t n = s.next_out - out_buf;
1096                         hashwrite(pack_file, out_buf, n);
1097                         pack_size += n;
1098                         s.next_out = out_buf;
1099                         s.avail_out = out_sz;
1100                 }
1101
1102                 switch (status) {
1103                 case Z_OK:
1104                 case Z_BUF_ERROR:
1105                 case Z_STREAM_END:
1106                         continue;
1107                 default:
1108                         die("unexpected deflate failure: %d", status);
1109                 }
1110         }
1111         git_deflate_end(&s);
1112         the_hash_algo->final_fn(oid.hash, &c);
1113
1114         if (oidout)
1115                 oidcpy(oidout, &oid);
1116
1117         e = insert_object(&oid);
1118
1119         if (mark)
1120                 insert_mark(mark, e);
1121
1122         if (e->idx.offset) {
1123                 duplicate_count_by_type[OBJ_BLOB]++;
1124                 truncate_pack(&checkpoint);
1125
1126         } else if (find_sha1_pack(oid.hash,
1127                                   get_all_packs(the_repository))) {
1128                 e->type = OBJ_BLOB;
1129                 e->pack_id = MAX_PACK_ID;
1130                 e->idx.offset = 1; /* just not zero! */
1131                 duplicate_count_by_type[OBJ_BLOB]++;
1132                 truncate_pack(&checkpoint);
1133
1134         } else {
1135                 e->depth = 0;
1136                 e->type = OBJ_BLOB;
1137                 e->pack_id = pack_id;
1138                 e->idx.offset = offset;
1139                 e->idx.crc32 = crc32_end(pack_file);
1140                 object_count++;
1141                 object_count_by_type[OBJ_BLOB]++;
1142         }
1143
1144         free(in_buf);
1145         free(out_buf);
1146 }
1147
1148 /* All calls must be guarded by find_object() or find_mark() to
1149  * ensure the 'struct object_entry' passed was written by this
1150  * process instance.  We unpack the entry by the offset, avoiding
1151  * the need for the corresponding .idx file.  This unpacking rule
1152  * works because we only use OBJ_REF_DELTA within the packfiles
1153  * created by fast-import.
1154  *
1155  * oe must not be NULL.  Such an oe usually comes from giving
1156  * an unknown SHA-1 to find_object() or an undefined mark to
1157  * find_mark().  Callers must test for this condition and use
1158  * the standard read_sha1_file() when it happens.
1159  *
1160  * oe->pack_id must not be MAX_PACK_ID.  Such an oe is usually from
1161  * find_mark(), where the mark was reloaded from an existing marks
1162  * file and is referencing an object that this fast-import process
1163  * instance did not write out to a packfile.  Callers must test for
1164  * this condition and use read_sha1_file() instead.
1165  */
1166 static void *gfi_unpack_entry(
1167         struct object_entry *oe,
1168         unsigned long *sizep)
1169 {
1170         enum object_type type;
1171         struct packed_git *p = all_packs[oe->pack_id];
1172         if (p == pack_data && p->pack_size < (pack_size + the_hash_algo->rawsz)) {
1173                 /* The object is stored in the packfile we are writing to
1174                  * and we have modified it since the last time we scanned
1175                  * back to read a previously written object.  If an old
1176                  * window covered [p->pack_size, p->pack_size + rawsz) its
1177                  * data is stale and is not valid.  Closing all windows
1178                  * and updating the packfile length ensures we can read
1179                  * the newly written data.
1180                  */
1181                 close_pack_windows(p);
1182                 hashflush(pack_file);
1183
1184                 /* We have to offer rawsz bytes additional on the end of
1185                  * the packfile as the core unpacker code assumes the
1186                  * footer is present at the file end and must promise
1187                  * at least rawsz bytes within any window it maps.  But
1188                  * we don't actually create the footer here.
1189                  */
1190                 p->pack_size = pack_size + the_hash_algo->rawsz;
1191         }
1192         return unpack_entry(the_repository, p, oe->idx.offset, &type, sizep);
1193 }
1194
1195 static const char *get_mode(const char *str, uint16_t *modep)
1196 {
1197         unsigned char c;
1198         uint16_t mode = 0;
1199
1200         while ((c = *str++) != ' ') {
1201                 if (c < '0' || c > '7')
1202                         return NULL;
1203                 mode = (mode << 3) + (c - '0');
1204         }
1205         *modep = mode;
1206         return str;
1207 }
1208
1209 static void load_tree(struct tree_entry *root)
1210 {
1211         struct object_id *oid = &root->versions[1].oid;
1212         struct object_entry *myoe;
1213         struct tree_content *t;
1214         unsigned long size;
1215         char *buf;
1216         const char *c;
1217
1218         root->tree = t = new_tree_content(8);
1219         if (is_null_oid(oid))
1220                 return;
1221
1222         myoe = find_object(oid);
1223         if (myoe && myoe->pack_id != MAX_PACK_ID) {
1224                 if (myoe->type != OBJ_TREE)
1225                         die("Not a tree: %s", oid_to_hex(oid));
1226                 t->delta_depth = myoe->depth;
1227                 buf = gfi_unpack_entry(myoe, &size);
1228                 if (!buf)
1229                         die("Can't load tree %s", oid_to_hex(oid));
1230         } else {
1231                 enum object_type type;
1232                 buf = read_object_file(oid, &type, &size);
1233                 if (!buf || type != OBJ_TREE)
1234                         die("Can't load tree %s", oid_to_hex(oid));
1235         }
1236
1237         c = buf;
1238         while (c != (buf + size)) {
1239                 struct tree_entry *e = new_tree_entry();
1240
1241                 if (t->entry_count == t->entry_capacity)
1242                         root->tree = t = grow_tree_content(t, t->entry_count);
1243                 t->entries[t->entry_count++] = e;
1244
1245                 e->tree = NULL;
1246                 c = get_mode(c, &e->versions[1].mode);
1247                 if (!c)
1248                         die("Corrupt mode in %s", oid_to_hex(oid));
1249                 e->versions[0].mode = e->versions[1].mode;
1250                 e->name = to_atom(c, strlen(c));
1251                 c += e->name->str_len + 1;
1252                 hashcpy(e->versions[0].oid.hash, (unsigned char *)c);
1253                 hashcpy(e->versions[1].oid.hash, (unsigned char *)c);
1254                 c += the_hash_algo->rawsz;
1255         }
1256         free(buf);
1257 }
1258
1259 static int tecmp0 (const void *_a, const void *_b)
1260 {
1261         struct tree_entry *a = *((struct tree_entry**)_a);
1262         struct tree_entry *b = *((struct tree_entry**)_b);
1263         return base_name_compare(
1264                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1265                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1266 }
1267
1268 static int tecmp1 (const void *_a, const void *_b)
1269 {
1270         struct tree_entry *a = *((struct tree_entry**)_a);
1271         struct tree_entry *b = *((struct tree_entry**)_b);
1272         return base_name_compare(
1273                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1274                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1275 }
1276
1277 static void mktree(struct tree_content *t, int v, struct strbuf *b)
1278 {
1279         size_t maxlen = 0;
1280         unsigned int i;
1281
1282         if (!v)
1283                 QSORT(t->entries, t->entry_count, tecmp0);
1284         else
1285                 QSORT(t->entries, t->entry_count, tecmp1);
1286
1287         for (i = 0; i < t->entry_count; i++) {
1288                 if (t->entries[i]->versions[v].mode)
1289                         maxlen += t->entries[i]->name->str_len + 34;
1290         }
1291
1292         strbuf_reset(b);
1293         strbuf_grow(b, maxlen);
1294         for (i = 0; i < t->entry_count; i++) {
1295                 struct tree_entry *e = t->entries[i];
1296                 if (!e->versions[v].mode)
1297                         continue;
1298                 strbuf_addf(b, "%o %s%c",
1299                         (unsigned int)(e->versions[v].mode & ~NO_DELTA),
1300                         e->name->str_dat, '\0');
1301                 strbuf_add(b, e->versions[v].oid.hash, the_hash_algo->rawsz);
1302         }
1303 }
1304
1305 static void store_tree(struct tree_entry *root)
1306 {
1307         struct tree_content *t;
1308         unsigned int i, j, del;
1309         struct last_object lo = { STRBUF_INIT, 0, 0, /* no_swap */ 1 };
1310         struct object_entry *le = NULL;
1311
1312         if (!is_null_oid(&root->versions[1].oid))
1313                 return;
1314
1315         if (!root->tree)
1316                 load_tree(root);
1317         t = root->tree;
1318
1319         for (i = 0; i < t->entry_count; i++) {
1320                 if (t->entries[i]->tree)
1321                         store_tree(t->entries[i]);
1322         }
1323
1324         if (!(root->versions[0].mode & NO_DELTA))
1325                 le = find_object(&root->versions[0].oid);
1326         if (S_ISDIR(root->versions[0].mode) && le && le->pack_id == pack_id) {
1327                 mktree(t, 0, &old_tree);
1328                 lo.data = old_tree;
1329                 lo.offset = le->idx.offset;
1330                 lo.depth = t->delta_depth;
1331         }
1332
1333         mktree(t, 1, &new_tree);
1334         store_object(OBJ_TREE, &new_tree, &lo, &root->versions[1].oid, 0);
1335
1336         t->delta_depth = lo.depth;
1337         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1338                 struct tree_entry *e = t->entries[i];
1339                 if (e->versions[1].mode) {
1340                         e->versions[0].mode = e->versions[1].mode;
1341                         oidcpy(&e->versions[0].oid, &e->versions[1].oid);
1342                         t->entries[j++] = e;
1343                 } else {
1344                         release_tree_entry(e);
1345                         del++;
1346                 }
1347         }
1348         t->entry_count -= del;
1349 }
1350
1351 static void tree_content_replace(
1352         struct tree_entry *root,
1353         const struct object_id *oid,
1354         const uint16_t mode,
1355         struct tree_content *newtree)
1356 {
1357         if (!S_ISDIR(mode))
1358                 die("Root cannot be a non-directory");
1359         oidclr(&root->versions[0].oid);
1360         oidcpy(&root->versions[1].oid, oid);
1361         if (root->tree)
1362                 release_tree_content_recursive(root->tree);
1363         root->tree = newtree;
1364 }
1365
1366 static int tree_content_set(
1367         struct tree_entry *root,
1368         const char *p,
1369         const struct object_id *oid,
1370         const uint16_t mode,
1371         struct tree_content *subtree)
1372 {
1373         struct tree_content *t;
1374         const char *slash1;
1375         unsigned int i, n;
1376         struct tree_entry *e;
1377
1378         slash1 = strchrnul(p, '/');
1379         n = slash1 - p;
1380         if (!n)
1381                 die("Empty path component found in input");
1382         if (!*slash1 && !S_ISDIR(mode) && subtree)
1383                 die("Non-directories cannot have subtrees");
1384
1385         if (!root->tree)
1386                 load_tree(root);
1387         t = root->tree;
1388         for (i = 0; i < t->entry_count; i++) {
1389                 e = t->entries[i];
1390                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1391                         if (!*slash1) {
1392                                 if (!S_ISDIR(mode)
1393                                                 && e->versions[1].mode == mode
1394                                                 && oideq(&e->versions[1].oid, oid))
1395                                         return 0;
1396                                 e->versions[1].mode = mode;
1397                                 oidcpy(&e->versions[1].oid, oid);
1398                                 if (e->tree)
1399                                         release_tree_content_recursive(e->tree);
1400                                 e->tree = subtree;
1401
1402                                 /*
1403                                  * We need to leave e->versions[0].sha1 alone
1404                                  * to avoid modifying the preimage tree used
1405                                  * when writing out the parent directory.
1406                                  * But after replacing the subdir with a
1407                                  * completely different one, it's not a good
1408                                  * delta base any more, and besides, we've
1409                                  * thrown away the tree entries needed to
1410                                  * make a delta against it.
1411                                  *
1412                                  * So let's just explicitly disable deltas
1413                                  * for the subtree.
1414                                  */
1415                                 if (S_ISDIR(e->versions[0].mode))
1416                                         e->versions[0].mode |= NO_DELTA;
1417
1418                                 oidclr(&root->versions[1].oid);
1419                                 return 1;
1420                         }
1421                         if (!S_ISDIR(e->versions[1].mode)) {
1422                                 e->tree = new_tree_content(8);
1423                                 e->versions[1].mode = S_IFDIR;
1424                         }
1425                         if (!e->tree)
1426                                 load_tree(e);
1427                         if (tree_content_set(e, slash1 + 1, oid, mode, subtree)) {
1428                                 oidclr(&root->versions[1].oid);
1429                                 return 1;
1430                         }
1431                         return 0;
1432                 }
1433         }
1434
1435         if (t->entry_count == t->entry_capacity)
1436                 root->tree = t = grow_tree_content(t, t->entry_count);
1437         e = new_tree_entry();
1438         e->name = to_atom(p, n);
1439         e->versions[0].mode = 0;
1440         oidclr(&e->versions[0].oid);
1441         t->entries[t->entry_count++] = e;
1442         if (*slash1) {
1443                 e->tree = new_tree_content(8);
1444                 e->versions[1].mode = S_IFDIR;
1445                 tree_content_set(e, slash1 + 1, oid, mode, subtree);
1446         } else {
1447                 e->tree = subtree;
1448                 e->versions[1].mode = mode;
1449                 oidcpy(&e->versions[1].oid, oid);
1450         }
1451         oidclr(&root->versions[1].oid);
1452         return 1;
1453 }
1454
1455 static int tree_content_remove(
1456         struct tree_entry *root,
1457         const char *p,
1458         struct tree_entry *backup_leaf,
1459         int allow_root)
1460 {
1461         struct tree_content *t;
1462         const char *slash1;
1463         unsigned int i, n;
1464         struct tree_entry *e;
1465
1466         slash1 = strchrnul(p, '/');
1467         n = slash1 - p;
1468
1469         if (!root->tree)
1470                 load_tree(root);
1471
1472         if (!*p && allow_root) {
1473                 e = root;
1474                 goto del_entry;
1475         }
1476
1477         t = root->tree;
1478         for (i = 0; i < t->entry_count; i++) {
1479                 e = t->entries[i];
1480                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1481                         if (*slash1 && !S_ISDIR(e->versions[1].mode))
1482                                 /*
1483                                  * If p names a file in some subdirectory, and a
1484                                  * file or symlink matching the name of the
1485                                  * parent directory of p exists, then p cannot
1486                                  * exist and need not be deleted.
1487                                  */
1488                                 return 1;
1489                         if (!*slash1 || !S_ISDIR(e->versions[1].mode))
1490                                 goto del_entry;
1491                         if (!e->tree)
1492                                 load_tree(e);
1493                         if (tree_content_remove(e, slash1 + 1, backup_leaf, 0)) {
1494                                 for (n = 0; n < e->tree->entry_count; n++) {
1495                                         if (e->tree->entries[n]->versions[1].mode) {
1496                                                 oidclr(&root->versions[1].oid);
1497                                                 return 1;
1498                                         }
1499                                 }
1500                                 backup_leaf = NULL;
1501                                 goto del_entry;
1502                         }
1503                         return 0;
1504                 }
1505         }
1506         return 0;
1507
1508 del_entry:
1509         if (backup_leaf)
1510                 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1511         else if (e->tree)
1512                 release_tree_content_recursive(e->tree);
1513         e->tree = NULL;
1514         e->versions[1].mode = 0;
1515         oidclr(&e->versions[1].oid);
1516         oidclr(&root->versions[1].oid);
1517         return 1;
1518 }
1519
1520 static int tree_content_get(
1521         struct tree_entry *root,
1522         const char *p,
1523         struct tree_entry *leaf,
1524         int allow_root)
1525 {
1526         struct tree_content *t;
1527         const char *slash1;
1528         unsigned int i, n;
1529         struct tree_entry *e;
1530
1531         slash1 = strchrnul(p, '/');
1532         n = slash1 - p;
1533         if (!n && !allow_root)
1534                 die("Empty path component found in input");
1535
1536         if (!root->tree)
1537                 load_tree(root);
1538
1539         if (!n) {
1540                 e = root;
1541                 goto found_entry;
1542         }
1543
1544         t = root->tree;
1545         for (i = 0; i < t->entry_count; i++) {
1546                 e = t->entries[i];
1547                 if (e->name->str_len == n && !fspathncmp(p, e->name->str_dat, n)) {
1548                         if (!*slash1)
1549                                 goto found_entry;
1550                         if (!S_ISDIR(e->versions[1].mode))
1551                                 return 0;
1552                         if (!e->tree)
1553                                 load_tree(e);
1554                         return tree_content_get(e, slash1 + 1, leaf, 0);
1555                 }
1556         }
1557         return 0;
1558
1559 found_entry:
1560         memcpy(leaf, e, sizeof(*leaf));
1561         if (e->tree && is_null_oid(&e->versions[1].oid))
1562                 leaf->tree = dup_tree_content(e->tree);
1563         else
1564                 leaf->tree = NULL;
1565         return 1;
1566 }
1567
1568 static int update_branch(struct branch *b)
1569 {
1570         static const char *msg = "fast-import";
1571         struct ref_transaction *transaction;
1572         struct object_id old_oid;
1573         struct strbuf err = STRBUF_INIT;
1574
1575         if (is_null_oid(&b->oid)) {
1576                 if (b->delete)
1577                         delete_ref(NULL, b->name, NULL, 0);
1578                 return 0;
1579         }
1580         if (read_ref(b->name, &old_oid))
1581                 oidclr(&old_oid);
1582         if (!force_update && !is_null_oid(&old_oid)) {
1583                 struct commit *old_cmit, *new_cmit;
1584
1585                 old_cmit = lookup_commit_reference_gently(the_repository,
1586                                                           &old_oid, 0);
1587                 new_cmit = lookup_commit_reference_gently(the_repository,
1588                                                           &b->oid, 0);
1589                 if (!old_cmit || !new_cmit)
1590                         return error("Branch %s is missing commits.", b->name);
1591
1592                 if (!in_merge_bases(old_cmit, new_cmit)) {
1593                         warning("Not updating %s"
1594                                 " (new tip %s does not contain %s)",
1595                                 b->name, oid_to_hex(&b->oid),
1596                                 oid_to_hex(&old_oid));
1597                         return -1;
1598                 }
1599         }
1600         transaction = ref_transaction_begin(&err);
1601         if (!transaction ||
1602             ref_transaction_update(transaction, b->name, &b->oid, &old_oid,
1603                                    0, msg, &err) ||
1604             ref_transaction_commit(transaction, &err)) {
1605                 ref_transaction_free(transaction);
1606                 error("%s", err.buf);
1607                 strbuf_release(&err);
1608                 return -1;
1609         }
1610         ref_transaction_free(transaction);
1611         strbuf_release(&err);
1612         return 0;
1613 }
1614
1615 static void dump_branches(void)
1616 {
1617         unsigned int i;
1618         struct branch *b;
1619
1620         for (i = 0; i < branch_table_sz; i++) {
1621                 for (b = branch_table[i]; b; b = b->table_next_branch)
1622                         failure |= update_branch(b);
1623         }
1624 }
1625
1626 static void dump_tags(void)
1627 {
1628         static const char *msg = "fast-import";
1629         struct tag *t;
1630         struct strbuf ref_name = STRBUF_INIT;
1631         struct strbuf err = STRBUF_INIT;
1632         struct ref_transaction *transaction;
1633
1634         transaction = ref_transaction_begin(&err);
1635         if (!transaction) {
1636                 failure |= error("%s", err.buf);
1637                 goto cleanup;
1638         }
1639         for (t = first_tag; t; t = t->next_tag) {
1640                 strbuf_reset(&ref_name);
1641                 strbuf_addf(&ref_name, "refs/tags/%s", t->name);
1642
1643                 if (ref_transaction_update(transaction, ref_name.buf,
1644                                            &t->oid, NULL, 0, msg, &err)) {
1645                         failure |= error("%s", err.buf);
1646                         goto cleanup;
1647                 }
1648         }
1649         if (ref_transaction_commit(transaction, &err))
1650                 failure |= error("%s", err.buf);
1651
1652  cleanup:
1653         ref_transaction_free(transaction);
1654         strbuf_release(&ref_name);
1655         strbuf_release(&err);
1656 }
1657
1658 static void dump_marks_helper(FILE *f,
1659         uintmax_t base,
1660         struct mark_set *m)
1661 {
1662         uintmax_t k;
1663         if (m->shift) {
1664                 for (k = 0; k < 1024; k++) {
1665                         if (m->data.sets[k])
1666                                 dump_marks_helper(f, base + (k << m->shift),
1667                                         m->data.sets[k]);
1668                 }
1669         } else {
1670                 for (k = 0; k < 1024; k++) {
1671                         if (m->data.marked[k])
1672                                 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1673                                         oid_to_hex(&m->data.marked[k]->idx.oid));
1674                 }
1675         }
1676 }
1677
1678 static void dump_marks(void)
1679 {
1680         struct lock_file mark_lock = LOCK_INIT;
1681         FILE *f;
1682
1683         if (!export_marks_file || (import_marks_file && !import_marks_file_done))
1684                 return;
1685
1686         if (safe_create_leading_directories_const(export_marks_file)) {
1687                 failure |= error_errno("unable to create leading directories of %s",
1688                                        export_marks_file);
1689                 return;
1690         }
1691
1692         if (hold_lock_file_for_update(&mark_lock, export_marks_file, 0) < 0) {
1693                 failure |= error_errno("Unable to write marks file %s",
1694                                        export_marks_file);
1695                 return;
1696         }
1697
1698         f = fdopen_lock_file(&mark_lock, "w");
1699         if (!f) {
1700                 int saved_errno = errno;
1701                 rollback_lock_file(&mark_lock);
1702                 failure |= error("Unable to write marks file %s: %s",
1703                         export_marks_file, strerror(saved_errno));
1704                 return;
1705         }
1706
1707         dump_marks_helper(f, 0, marks);
1708         if (commit_lock_file(&mark_lock)) {
1709                 failure |= error_errno("Unable to write file %s",
1710                                        export_marks_file);
1711                 return;
1712         }
1713 }
1714
1715 static void read_marks(void)
1716 {
1717         char line[512];
1718         FILE *f = fopen(import_marks_file, "r");
1719         if (f)
1720                 ;
1721         else if (import_marks_file_ignore_missing && errno == ENOENT)
1722                 goto done; /* Marks file does not exist */
1723         else
1724                 die_errno("cannot read '%s'", import_marks_file);
1725         while (fgets(line, sizeof(line), f)) {
1726                 uintmax_t mark;
1727                 char *end;
1728                 struct object_id oid;
1729                 struct object_entry *e;
1730
1731                 end = strchr(line, '\n');
1732                 if (line[0] != ':' || !end)
1733                         die("corrupt mark line: %s", line);
1734                 *end = 0;
1735                 mark = strtoumax(line + 1, &end, 10);
1736                 if (!mark || end == line + 1
1737                         || *end != ' ' || get_oid_hex(end + 1, &oid))
1738                         die("corrupt mark line: %s", line);
1739                 e = find_object(&oid);
1740                 if (!e) {
1741                         enum object_type type = oid_object_info(the_repository,
1742                                                                 &oid, NULL);
1743                         if (type < 0)
1744                                 die("object not found: %s", oid_to_hex(&oid));
1745                         e = insert_object(&oid);
1746                         e->type = type;
1747                         e->pack_id = MAX_PACK_ID;
1748                         e->idx.offset = 1; /* just not zero! */
1749                 }
1750                 insert_mark(mark, e);
1751         }
1752         fclose(f);
1753 done:
1754         import_marks_file_done = 1;
1755 }
1756
1757
1758 static int read_next_command(void)
1759 {
1760         static int stdin_eof = 0;
1761
1762         if (stdin_eof) {
1763                 unread_command_buf = 0;
1764                 return EOF;
1765         }
1766
1767         for (;;) {
1768                 if (unread_command_buf) {
1769                         unread_command_buf = 0;
1770                 } else {
1771                         struct recent_command *rc;
1772
1773                         strbuf_detach(&command_buf, NULL);
1774                         stdin_eof = strbuf_getline_lf(&command_buf, stdin);
1775                         if (stdin_eof)
1776                                 return EOF;
1777
1778                         if (!seen_data_command
1779                                 && !starts_with(command_buf.buf, "feature ")
1780                                 && !starts_with(command_buf.buf, "option ")) {
1781                                 parse_argv();
1782                         }
1783
1784                         rc = rc_free;
1785                         if (rc)
1786                                 rc_free = rc->next;
1787                         else {
1788                                 rc = cmd_hist.next;
1789                                 cmd_hist.next = rc->next;
1790                                 cmd_hist.next->prev = &cmd_hist;
1791                                 free(rc->buf);
1792                         }
1793
1794                         rc->buf = command_buf.buf;
1795                         rc->prev = cmd_tail;
1796                         rc->next = cmd_hist.prev;
1797                         rc->prev->next = rc;
1798                         cmd_tail = rc;
1799                 }
1800                 if (command_buf.buf[0] == '#')
1801                         continue;
1802                 return 0;
1803         }
1804 }
1805
1806 static void skip_optional_lf(void)
1807 {
1808         int term_char = fgetc(stdin);
1809         if (term_char != '\n' && term_char != EOF)
1810                 ungetc(term_char, stdin);
1811 }
1812
1813 static void parse_mark(void)
1814 {
1815         const char *v;
1816         if (skip_prefix(command_buf.buf, "mark :", &v)) {
1817                 next_mark = strtoumax(v, NULL, 10);
1818                 read_next_command();
1819         }
1820         else
1821                 next_mark = 0;
1822 }
1823
1824 static void parse_original_identifier(void)
1825 {
1826         const char *v;
1827         if (skip_prefix(command_buf.buf, "original-oid ", &v))
1828                 read_next_command();
1829 }
1830
1831 static int parse_data(struct strbuf *sb, uintmax_t limit, uintmax_t *len_res)
1832 {
1833         const char *data;
1834         strbuf_reset(sb);
1835
1836         if (!skip_prefix(command_buf.buf, "data ", &data))
1837                 die("Expected 'data n' command, found: %s", command_buf.buf);
1838
1839         if (skip_prefix(data, "<<", &data)) {
1840                 char *term = xstrdup(data);
1841                 size_t term_len = command_buf.len - (data - command_buf.buf);
1842
1843                 strbuf_detach(&command_buf, NULL);
1844                 for (;;) {
1845                         if (strbuf_getline_lf(&command_buf, stdin) == EOF)
1846                                 die("EOF in data (terminator '%s' not found)", term);
1847                         if (term_len == command_buf.len
1848                                 && !strcmp(term, command_buf.buf))
1849                                 break;
1850                         strbuf_addbuf(sb, &command_buf);
1851                         strbuf_addch(sb, '\n');
1852                 }
1853                 free(term);
1854         }
1855         else {
1856                 uintmax_t len = strtoumax(data, NULL, 10);
1857                 size_t n = 0, length = (size_t)len;
1858
1859                 if (limit && limit < len) {
1860                         *len_res = len;
1861                         return 0;
1862                 }
1863                 if (length < len)
1864                         die("data is too large to use in this context");
1865
1866                 while (n < length) {
1867                         size_t s = strbuf_fread(sb, length - n, stdin);
1868                         if (!s && feof(stdin))
1869                                 die("EOF in data (%lu bytes remaining)",
1870                                         (unsigned long)(length - n));
1871                         n += s;
1872                 }
1873         }
1874
1875         skip_optional_lf();
1876         return 1;
1877 }
1878
1879 static int validate_raw_date(const char *src, struct strbuf *result)
1880 {
1881         const char *orig_src = src;
1882         char *endp;
1883         unsigned long num;
1884
1885         errno = 0;
1886
1887         num = strtoul(src, &endp, 10);
1888         /* NEEDSWORK: perhaps check for reasonable values? */
1889         if (errno || endp == src || *endp != ' ')
1890                 return -1;
1891
1892         src = endp + 1;
1893         if (*src != '-' && *src != '+')
1894                 return -1;
1895
1896         num = strtoul(src + 1, &endp, 10);
1897         if (errno || endp == src + 1 || *endp || 1400 < num)
1898                 return -1;
1899
1900         strbuf_addstr(result, orig_src);
1901         return 0;
1902 }
1903
1904 static char *parse_ident(const char *buf)
1905 {
1906         const char *ltgt;
1907         size_t name_len;
1908         struct strbuf ident = STRBUF_INIT;
1909
1910         /* ensure there is a space delimiter even if there is no name */
1911         if (*buf == '<')
1912                 --buf;
1913
1914         ltgt = buf + strcspn(buf, "<>");
1915         if (*ltgt != '<')
1916                 die("Missing < in ident string: %s", buf);
1917         if (ltgt != buf && ltgt[-1] != ' ')
1918                 die("Missing space before < in ident string: %s", buf);
1919         ltgt = ltgt + 1 + strcspn(ltgt + 1, "<>");
1920         if (*ltgt != '>')
1921                 die("Missing > in ident string: %s", buf);
1922         ltgt++;
1923         if (*ltgt != ' ')
1924                 die("Missing space after > in ident string: %s", buf);
1925         ltgt++;
1926         name_len = ltgt - buf;
1927         strbuf_add(&ident, buf, name_len);
1928
1929         switch (whenspec) {
1930         case WHENSPEC_RAW:
1931                 if (validate_raw_date(ltgt, &ident) < 0)
1932                         die("Invalid raw date \"%s\" in ident: %s", ltgt, buf);
1933                 break;
1934         case WHENSPEC_RFC2822:
1935                 if (parse_date(ltgt, &ident) < 0)
1936                         die("Invalid rfc2822 date \"%s\" in ident: %s", ltgt, buf);
1937                 break;
1938         case WHENSPEC_NOW:
1939                 if (strcmp("now", ltgt))
1940                         die("Date in ident must be 'now': %s", buf);
1941                 datestamp(&ident);
1942                 break;
1943         }
1944
1945         return strbuf_detach(&ident, NULL);
1946 }
1947
1948 static void parse_and_store_blob(
1949         struct last_object *last,
1950         struct object_id *oidout,
1951         uintmax_t mark)
1952 {
1953         static struct strbuf buf = STRBUF_INIT;
1954         uintmax_t len;
1955
1956         if (parse_data(&buf, big_file_threshold, &len))
1957                 store_object(OBJ_BLOB, &buf, last, oidout, mark);
1958         else {
1959                 if (last) {
1960                         strbuf_release(&last->data);
1961                         last->offset = 0;
1962                         last->depth = 0;
1963                 }
1964                 stream_blob(len, oidout, mark);
1965                 skip_optional_lf();
1966         }
1967 }
1968
1969 static void parse_new_blob(void)
1970 {
1971         read_next_command();
1972         parse_mark();
1973         parse_original_identifier();
1974         parse_and_store_blob(&last_blob, NULL, next_mark);
1975 }
1976
1977 static void unload_one_branch(void)
1978 {
1979         while (cur_active_branches
1980                 && cur_active_branches >= max_active_branches) {
1981                 uintmax_t min_commit = ULONG_MAX;
1982                 struct branch *e, *l = NULL, *p = NULL;
1983
1984                 for (e = active_branches; e; e = e->active_next_branch) {
1985                         if (e->last_commit < min_commit) {
1986                                 p = l;
1987                                 min_commit = e->last_commit;
1988                         }
1989                         l = e;
1990                 }
1991
1992                 if (p) {
1993                         e = p->active_next_branch;
1994                         p->active_next_branch = e->active_next_branch;
1995                 } else {
1996                         e = active_branches;
1997                         active_branches = e->active_next_branch;
1998                 }
1999                 e->active = 0;
2000                 e->active_next_branch = NULL;
2001                 if (e->branch_tree.tree) {
2002                         release_tree_content_recursive(e->branch_tree.tree);
2003                         e->branch_tree.tree = NULL;
2004                 }
2005                 cur_active_branches--;
2006         }
2007 }
2008
2009 static void load_branch(struct branch *b)
2010 {
2011         load_tree(&b->branch_tree);
2012         if (!b->active) {
2013                 b->active = 1;
2014                 b->active_next_branch = active_branches;
2015                 active_branches = b;
2016                 cur_active_branches++;
2017                 branch_load_count++;
2018         }
2019 }
2020
2021 static unsigned char convert_num_notes_to_fanout(uintmax_t num_notes)
2022 {
2023         unsigned char fanout = 0;
2024         while ((num_notes >>= 8))
2025                 fanout++;
2026         return fanout;
2027 }
2028
2029 static void construct_path_with_fanout(const char *hex_sha1,
2030                 unsigned char fanout, char *path)
2031 {
2032         unsigned int i = 0, j = 0;
2033         if (fanout >= the_hash_algo->rawsz)
2034                 die("Too large fanout (%u)", fanout);
2035         while (fanout) {
2036                 path[i++] = hex_sha1[j++];
2037                 path[i++] = hex_sha1[j++];
2038                 path[i++] = '/';
2039                 fanout--;
2040         }
2041         memcpy(path + i, hex_sha1 + j, the_hash_algo->hexsz - j);
2042         path[i + the_hash_algo->hexsz - j] = '\0';
2043 }
2044
2045 static uintmax_t do_change_note_fanout(
2046                 struct tree_entry *orig_root, struct tree_entry *root,
2047                 char *hex_oid, unsigned int hex_oid_len,
2048                 char *fullpath, unsigned int fullpath_len,
2049                 unsigned char fanout)
2050 {
2051         struct tree_content *t;
2052         struct tree_entry *e, leaf;
2053         unsigned int i, tmp_hex_oid_len, tmp_fullpath_len;
2054         uintmax_t num_notes = 0;
2055         struct object_id oid;
2056         /* hex oid + '/' between each pair of hex digits + NUL */
2057         char realpath[GIT_MAX_HEXSZ + ((GIT_MAX_HEXSZ / 2) - 1) + 1];
2058         const unsigned hexsz = the_hash_algo->hexsz;
2059
2060         if (!root->tree)
2061                 load_tree(root);
2062         t = root->tree;
2063
2064         for (i = 0; t && i < t->entry_count; i++) {
2065                 e = t->entries[i];
2066                 tmp_hex_oid_len = hex_oid_len + e->name->str_len;
2067                 tmp_fullpath_len = fullpath_len;
2068
2069                 /*
2070                  * We're interested in EITHER existing note entries (entries
2071                  * with exactly 40 hex chars in path, not including directory
2072                  * separators), OR directory entries that may contain note
2073                  * entries (with < 40 hex chars in path).
2074                  * Also, each path component in a note entry must be a multiple
2075                  * of 2 chars.
2076                  */
2077                 if (!e->versions[1].mode ||
2078                     tmp_hex_oid_len > hexsz ||
2079                     e->name->str_len % 2)
2080                         continue;
2081
2082                 /* This _may_ be a note entry, or a subdir containing notes */
2083                 memcpy(hex_oid + hex_oid_len, e->name->str_dat,
2084                        e->name->str_len);
2085                 if (tmp_fullpath_len)
2086                         fullpath[tmp_fullpath_len++] = '/';
2087                 memcpy(fullpath + tmp_fullpath_len, e->name->str_dat,
2088                        e->name->str_len);
2089                 tmp_fullpath_len += e->name->str_len;
2090                 fullpath[tmp_fullpath_len] = '\0';
2091
2092                 if (tmp_hex_oid_len == hexsz && !get_oid_hex(hex_oid, &oid)) {
2093                         /* This is a note entry */
2094                         if (fanout == 0xff) {
2095                                 /* Counting mode, no rename */
2096                                 num_notes++;
2097                                 continue;
2098                         }
2099                         construct_path_with_fanout(hex_oid, fanout, realpath);
2100                         if (!strcmp(fullpath, realpath)) {
2101                                 /* Note entry is in correct location */
2102                                 num_notes++;
2103                                 continue;
2104                         }
2105
2106                         /* Rename fullpath to realpath */
2107                         if (!tree_content_remove(orig_root, fullpath, &leaf, 0))
2108                                 die("Failed to remove path %s", fullpath);
2109                         tree_content_set(orig_root, realpath,
2110                                 &leaf.versions[1].oid,
2111                                 leaf.versions[1].mode,
2112                                 leaf.tree);
2113                 } else if (S_ISDIR(e->versions[1].mode)) {
2114                         /* This is a subdir that may contain note entries */
2115                         num_notes += do_change_note_fanout(orig_root, e,
2116                                 hex_oid, tmp_hex_oid_len,
2117                                 fullpath, tmp_fullpath_len, fanout);
2118                 }
2119
2120                 /* The above may have reallocated the current tree_content */
2121                 t = root->tree;
2122         }
2123         return num_notes;
2124 }
2125
2126 static uintmax_t change_note_fanout(struct tree_entry *root,
2127                 unsigned char fanout)
2128 {
2129         /*
2130          * The size of path is due to one slash between every two hex digits,
2131          * plus the terminating NUL.  Note that there is no slash at the end, so
2132          * the number of slashes is one less than half the number of hex
2133          * characters.
2134          */
2135         char hex_oid[GIT_MAX_HEXSZ], path[GIT_MAX_HEXSZ + (GIT_MAX_HEXSZ / 2) - 1 + 1];
2136         return do_change_note_fanout(root, root, hex_oid, 0, path, 0, fanout);
2137 }
2138
2139 /*
2140  * Given a pointer into a string, parse a mark reference:
2141  *
2142  *   idnum ::= ':' bigint;
2143  *
2144  * Return the first character after the value in *endptr.
2145  *
2146  * Complain if the following character is not what is expected,
2147  * either a space or end of the string.
2148  */
2149 static uintmax_t parse_mark_ref(const char *p, char **endptr)
2150 {
2151         uintmax_t mark;
2152
2153         assert(*p == ':');
2154         p++;
2155         mark = strtoumax(p, endptr, 10);
2156         if (*endptr == p)
2157                 die("No value after ':' in mark: %s", command_buf.buf);
2158         return mark;
2159 }
2160
2161 /*
2162  * Parse the mark reference, and complain if this is not the end of
2163  * the string.
2164  */
2165 static uintmax_t parse_mark_ref_eol(const char *p)
2166 {
2167         char *end;
2168         uintmax_t mark;
2169
2170         mark = parse_mark_ref(p, &end);
2171         if (*end != '\0')
2172                 die("Garbage after mark: %s", command_buf.buf);
2173         return mark;
2174 }
2175
2176 /*
2177  * Parse the mark reference, demanding a trailing space.  Return a
2178  * pointer to the space.
2179  */
2180 static uintmax_t parse_mark_ref_space(const char **p)
2181 {
2182         uintmax_t mark;
2183         char *end;
2184
2185         mark = parse_mark_ref(*p, &end);
2186         if (*end++ != ' ')
2187                 die("Missing space after mark: %s", command_buf.buf);
2188         *p = end;
2189         return mark;
2190 }
2191
2192 static void file_change_m(const char *p, struct branch *b)
2193 {
2194         static struct strbuf uq = STRBUF_INIT;
2195         const char *endp;
2196         struct object_entry *oe;
2197         struct object_id oid;
2198         uint16_t mode, inline_data = 0;
2199
2200         p = get_mode(p, &mode);
2201         if (!p)
2202                 die("Corrupt mode: %s", command_buf.buf);
2203         switch (mode) {
2204         case 0644:
2205         case 0755:
2206                 mode |= S_IFREG;
2207         case S_IFREG | 0644:
2208         case S_IFREG | 0755:
2209         case S_IFLNK:
2210         case S_IFDIR:
2211         case S_IFGITLINK:
2212                 /* ok */
2213                 break;
2214         default:
2215                 die("Corrupt mode: %s", command_buf.buf);
2216         }
2217
2218         if (*p == ':') {
2219                 oe = find_mark(parse_mark_ref_space(&p));
2220                 oidcpy(&oid, &oe->idx.oid);
2221         } else if (skip_prefix(p, "inline ", &p)) {
2222                 inline_data = 1;
2223                 oe = NULL; /* not used with inline_data, but makes gcc happy */
2224         } else {
2225                 if (parse_oid_hex(p, &oid, &p))
2226                         die("Invalid dataref: %s", command_buf.buf);
2227                 oe = find_object(&oid);
2228                 if (*p++ != ' ')
2229                         die("Missing space after SHA1: %s", command_buf.buf);
2230         }
2231
2232         strbuf_reset(&uq);
2233         if (!unquote_c_style(&uq, p, &endp)) {
2234                 if (*endp)
2235                         die("Garbage after path in: %s", command_buf.buf);
2236                 p = uq.buf;
2237         }
2238
2239         /* Git does not track empty, non-toplevel directories. */
2240         if (S_ISDIR(mode) && is_empty_tree_oid(&oid) && *p) {
2241                 tree_content_remove(&b->branch_tree, p, NULL, 0);
2242                 return;
2243         }
2244
2245         if (S_ISGITLINK(mode)) {
2246                 if (inline_data)
2247                         die("Git links cannot be specified 'inline': %s",
2248                                 command_buf.buf);
2249                 else if (oe) {
2250                         if (oe->type != OBJ_COMMIT)
2251                                 die("Not a commit (actually a %s): %s",
2252                                         type_name(oe->type), command_buf.buf);
2253                 }
2254                 /*
2255                  * Accept the sha1 without checking; it expected to be in
2256                  * another repository.
2257                  */
2258         } else if (inline_data) {
2259                 if (S_ISDIR(mode))
2260                         die("Directories cannot be specified 'inline': %s",
2261                                 command_buf.buf);
2262                 if (p != uq.buf) {
2263                         strbuf_addstr(&uq, p);
2264                         p = uq.buf;
2265                 }
2266                 while (read_next_command() != EOF) {
2267                         const char *v;
2268                         if (skip_prefix(command_buf.buf, "cat-blob ", &v))
2269                                 parse_cat_blob(v);
2270                         else {
2271                                 parse_and_store_blob(&last_blob, &oid, 0);
2272                                 break;
2273                         }
2274                 }
2275         } else {
2276                 enum object_type expected = S_ISDIR(mode) ?
2277                                                 OBJ_TREE: OBJ_BLOB;
2278                 enum object_type type = oe ? oe->type :
2279                                         oid_object_info(the_repository, &oid,
2280                                                         NULL);
2281                 if (type < 0)
2282                         die("%s not found: %s",
2283                                         S_ISDIR(mode) ?  "Tree" : "Blob",
2284                                         command_buf.buf);
2285                 if (type != expected)
2286                         die("Not a %s (actually a %s): %s",
2287                                 type_name(expected), type_name(type),
2288                                 command_buf.buf);
2289         }
2290
2291         if (!*p) {
2292                 tree_content_replace(&b->branch_tree, &oid, mode, NULL);
2293                 return;
2294         }
2295         tree_content_set(&b->branch_tree, p, &oid, mode, NULL);
2296 }
2297
2298 static void file_change_d(const char *p, struct branch *b)
2299 {
2300         static struct strbuf uq = STRBUF_INIT;
2301         const char *endp;
2302
2303         strbuf_reset(&uq);
2304         if (!unquote_c_style(&uq, p, &endp)) {
2305                 if (*endp)
2306                         die("Garbage after path in: %s", command_buf.buf);
2307                 p = uq.buf;
2308         }
2309         tree_content_remove(&b->branch_tree, p, NULL, 1);
2310 }
2311
2312 static void file_change_cr(const char *s, struct branch *b, int rename)
2313 {
2314         const char *d;
2315         static struct strbuf s_uq = STRBUF_INIT;
2316         static struct strbuf d_uq = STRBUF_INIT;
2317         const char *endp;
2318         struct tree_entry leaf;
2319
2320         strbuf_reset(&s_uq);
2321         if (!unquote_c_style(&s_uq, s, &endp)) {
2322                 if (*endp != ' ')
2323                         die("Missing space after source: %s", command_buf.buf);
2324         } else {
2325                 endp = strchr(s, ' ');
2326                 if (!endp)
2327                         die("Missing space after source: %s", command_buf.buf);
2328                 strbuf_add(&s_uq, s, endp - s);
2329         }
2330         s = s_uq.buf;
2331
2332         endp++;
2333         if (!*endp)
2334                 die("Missing dest: %s", command_buf.buf);
2335
2336         d = endp;
2337         strbuf_reset(&d_uq);
2338         if (!unquote_c_style(&d_uq, d, &endp)) {
2339                 if (*endp)
2340                         die("Garbage after dest in: %s", command_buf.buf);
2341                 d = d_uq.buf;
2342         }
2343
2344         memset(&leaf, 0, sizeof(leaf));
2345         if (rename)
2346                 tree_content_remove(&b->branch_tree, s, &leaf, 1);
2347         else
2348                 tree_content_get(&b->branch_tree, s, &leaf, 1);
2349         if (!leaf.versions[1].mode)
2350                 die("Path %s not in branch", s);
2351         if (!*d) {      /* C "path/to/subdir" "" */
2352                 tree_content_replace(&b->branch_tree,
2353                         &leaf.versions[1].oid,
2354                         leaf.versions[1].mode,
2355                         leaf.tree);
2356                 return;
2357         }
2358         tree_content_set(&b->branch_tree, d,
2359                 &leaf.versions[1].oid,
2360                 leaf.versions[1].mode,
2361                 leaf.tree);
2362 }
2363
2364 static void note_change_n(const char *p, struct branch *b, unsigned char *old_fanout)
2365 {
2366         static struct strbuf uq = STRBUF_INIT;
2367         struct object_entry *oe;
2368         struct branch *s;
2369         struct object_id oid, commit_oid;
2370         char path[GIT_MAX_RAWSZ * 3];
2371         uint16_t inline_data = 0;
2372         unsigned char new_fanout;
2373
2374         /*
2375          * When loading a branch, we don't traverse its tree to count the real
2376          * number of notes (too expensive to do this for all non-note refs).
2377          * This means that recently loaded notes refs might incorrectly have
2378          * b->num_notes == 0, and consequently, old_fanout might be wrong.
2379          *
2380          * Fix this by traversing the tree and counting the number of notes
2381          * when b->num_notes == 0. If the notes tree is truly empty, the
2382          * calculation should not take long.
2383          */
2384         if (b->num_notes == 0 && *old_fanout == 0) {
2385                 /* Invoke change_note_fanout() in "counting mode". */
2386                 b->num_notes = change_note_fanout(&b->branch_tree, 0xff);
2387                 *old_fanout = convert_num_notes_to_fanout(b->num_notes);
2388         }
2389
2390         /* Now parse the notemodify command. */
2391         /* <dataref> or 'inline' */
2392         if (*p == ':') {
2393                 oe = find_mark(parse_mark_ref_space(&p));
2394                 oidcpy(&oid, &oe->idx.oid);
2395         } else if (skip_prefix(p, "inline ", &p)) {
2396                 inline_data = 1;
2397                 oe = NULL; /* not used with inline_data, but makes gcc happy */
2398         } else {
2399                 if (parse_oid_hex(p, &oid, &p))
2400                         die("Invalid dataref: %s", command_buf.buf);
2401                 oe = find_object(&oid);
2402                 if (*p++ != ' ')
2403                         die("Missing space after SHA1: %s", command_buf.buf);
2404         }
2405
2406         /* <commit-ish> */
2407         s = lookup_branch(p);
2408         if (s) {
2409                 if (is_null_oid(&s->oid))
2410                         die("Can't add a note on empty branch.");
2411                 oidcpy(&commit_oid, &s->oid);
2412         } else if (*p == ':') {
2413                 uintmax_t commit_mark = parse_mark_ref_eol(p);
2414                 struct object_entry *commit_oe = find_mark(commit_mark);
2415                 if (commit_oe->type != OBJ_COMMIT)
2416                         die("Mark :%" PRIuMAX " not a commit", commit_mark);
2417                 oidcpy(&commit_oid, &commit_oe->idx.oid);
2418         } else if (!get_oid(p, &commit_oid)) {
2419                 unsigned long size;
2420                 char *buf = read_object_with_reference(the_repository,
2421                                                        &commit_oid,
2422                                                        commit_type, &size,
2423                                                        &commit_oid);
2424                 if (!buf || size < the_hash_algo->hexsz + 6)
2425                         die("Not a valid commit: %s", p);
2426                 free(buf);
2427         } else
2428                 die("Invalid ref name or SHA1 expression: %s", p);
2429
2430         if (inline_data) {
2431                 if (p != uq.buf) {
2432                         strbuf_addstr(&uq, p);
2433                         p = uq.buf;
2434                 }
2435                 read_next_command();
2436                 parse_and_store_blob(&last_blob, &oid, 0);
2437         } else if (oe) {
2438                 if (oe->type != OBJ_BLOB)
2439                         die("Not a blob (actually a %s): %s",
2440                                 type_name(oe->type), command_buf.buf);
2441         } else if (!is_null_oid(&oid)) {
2442                 enum object_type type = oid_object_info(the_repository, &oid,
2443                                                         NULL);
2444                 if (type < 0)
2445                         die("Blob not found: %s", command_buf.buf);
2446                 if (type != OBJ_BLOB)
2447                         die("Not a blob (actually a %s): %s",
2448                             type_name(type), command_buf.buf);
2449         }
2450
2451         construct_path_with_fanout(oid_to_hex(&commit_oid), *old_fanout, path);
2452         if (tree_content_remove(&b->branch_tree, path, NULL, 0))
2453                 b->num_notes--;
2454
2455         if (is_null_oid(&oid))
2456                 return; /* nothing to insert */
2457
2458         b->num_notes++;
2459         new_fanout = convert_num_notes_to_fanout(b->num_notes);
2460         construct_path_with_fanout(oid_to_hex(&commit_oid), new_fanout, path);
2461         tree_content_set(&b->branch_tree, path, &oid, S_IFREG | 0644, NULL);
2462 }
2463
2464 static void file_change_deleteall(struct branch *b)
2465 {
2466         release_tree_content_recursive(b->branch_tree.tree);
2467         oidclr(&b->branch_tree.versions[0].oid);
2468         oidclr(&b->branch_tree.versions[1].oid);
2469         load_tree(&b->branch_tree);
2470         b->num_notes = 0;
2471 }
2472
2473 static void parse_from_commit(struct branch *b, char *buf, unsigned long size)
2474 {
2475         if (!buf || size < the_hash_algo->hexsz + 6)
2476                 die("Not a valid commit: %s", oid_to_hex(&b->oid));
2477         if (memcmp("tree ", buf, 5)
2478                 || get_oid_hex(buf + 5, &b->branch_tree.versions[1].oid))
2479                 die("The commit %s is corrupt", oid_to_hex(&b->oid));
2480         oidcpy(&b->branch_tree.versions[0].oid,
2481                &b->branch_tree.versions[1].oid);
2482 }
2483
2484 static void parse_from_existing(struct branch *b)
2485 {
2486         if (is_null_oid(&b->oid)) {
2487                 oidclr(&b->branch_tree.versions[0].oid);
2488                 oidclr(&b->branch_tree.versions[1].oid);
2489         } else {
2490                 unsigned long size;
2491                 char *buf;
2492
2493                 buf = read_object_with_reference(the_repository,
2494                                                  &b->oid, commit_type, &size,
2495                                                  &b->oid);
2496                 parse_from_commit(b, buf, size);
2497                 free(buf);
2498         }
2499 }
2500
2501 static int parse_from(struct branch *b)
2502 {
2503         const char *from;
2504         struct branch *s;
2505         struct object_id oid;
2506
2507         if (!skip_prefix(command_buf.buf, "from ", &from))
2508                 return 0;
2509
2510         oidcpy(&oid, &b->branch_tree.versions[1].oid);
2511
2512         s = lookup_branch(from);
2513         if (b == s)
2514                 die("Can't create a branch from itself: %s", b->name);
2515         else if (s) {
2516                 struct object_id *t = &s->branch_tree.versions[1].oid;
2517                 oidcpy(&b->oid, &s->oid);
2518                 oidcpy(&b->branch_tree.versions[0].oid, t);
2519                 oidcpy(&b->branch_tree.versions[1].oid, t);
2520         } else if (*from == ':') {
2521                 uintmax_t idnum = parse_mark_ref_eol(from);
2522                 struct object_entry *oe = find_mark(idnum);
2523                 if (oe->type != OBJ_COMMIT)
2524                         die("Mark :%" PRIuMAX " not a commit", idnum);
2525                 if (!oideq(&b->oid, &oe->idx.oid)) {
2526                         oidcpy(&b->oid, &oe->idx.oid);
2527                         if (oe->pack_id != MAX_PACK_ID) {
2528                                 unsigned long size;
2529                                 char *buf = gfi_unpack_entry(oe, &size);
2530                                 parse_from_commit(b, buf, size);
2531                                 free(buf);
2532                         } else
2533                                 parse_from_existing(b);
2534                 }
2535         } else if (!get_oid(from, &b->oid)) {
2536                 parse_from_existing(b);
2537                 if (is_null_oid(&b->oid))
2538                         b->delete = 1;
2539         }
2540         else
2541                 die("Invalid ref name or SHA1 expression: %s", from);
2542
2543         if (b->branch_tree.tree && !oideq(&oid, &b->branch_tree.versions[1].oid)) {
2544                 release_tree_content_recursive(b->branch_tree.tree);
2545                 b->branch_tree.tree = NULL;
2546         }
2547
2548         read_next_command();
2549         return 1;
2550 }
2551
2552 static struct hash_list *parse_merge(unsigned int *count)
2553 {
2554         struct hash_list *list = NULL, **tail = &list, *n;
2555         const char *from;
2556         struct branch *s;
2557
2558         *count = 0;
2559         while (skip_prefix(command_buf.buf, "merge ", &from)) {
2560                 n = xmalloc(sizeof(*n));
2561                 s = lookup_branch(from);
2562                 if (s)
2563                         oidcpy(&n->oid, &s->oid);
2564                 else if (*from == ':') {
2565                         uintmax_t idnum = parse_mark_ref_eol(from);
2566                         struct object_entry *oe = find_mark(idnum);
2567                         if (oe->type != OBJ_COMMIT)
2568                                 die("Mark :%" PRIuMAX " not a commit", idnum);
2569                         oidcpy(&n->oid, &oe->idx.oid);
2570                 } else if (!get_oid(from, &n->oid)) {
2571                         unsigned long size;
2572                         char *buf = read_object_with_reference(the_repository,
2573                                                                &n->oid,
2574                                                                commit_type,
2575                                                                &size, &n->oid);
2576                         if (!buf || size < the_hash_algo->hexsz + 6)
2577                                 die("Not a valid commit: %s", from);
2578                         free(buf);
2579                 } else
2580                         die("Invalid ref name or SHA1 expression: %s", from);
2581
2582                 n->next = NULL;
2583                 *tail = n;
2584                 tail = &n->next;
2585
2586                 (*count)++;
2587                 read_next_command();
2588         }
2589         return list;
2590 }
2591
2592 static void parse_new_commit(const char *arg)
2593 {
2594         static struct strbuf msg = STRBUF_INIT;
2595         struct branch *b;
2596         char *author = NULL;
2597         char *committer = NULL;
2598         const char *encoding = NULL;
2599         struct hash_list *merge_list = NULL;
2600         unsigned int merge_count;
2601         unsigned char prev_fanout, new_fanout;
2602         const char *v;
2603
2604         b = lookup_branch(arg);
2605         if (!b)
2606                 b = new_branch(arg);
2607
2608         read_next_command();
2609         parse_mark();
2610         parse_original_identifier();
2611         if (skip_prefix(command_buf.buf, "author ", &v)) {
2612                 author = parse_ident(v);
2613                 read_next_command();
2614         }
2615         if (skip_prefix(command_buf.buf, "committer ", &v)) {
2616                 committer = parse_ident(v);
2617                 read_next_command();
2618         }
2619         if (!committer)
2620                 die("Expected committer but didn't get one");
2621         if (skip_prefix(command_buf.buf, "encoding ", &encoding))
2622                 read_next_command();
2623         parse_data(&msg, 0, NULL);
2624         read_next_command();
2625         parse_from(b);
2626         merge_list = parse_merge(&merge_count);
2627
2628         /* ensure the branch is active/loaded */
2629         if (!b->branch_tree.tree || !max_active_branches) {
2630                 unload_one_branch();
2631                 load_branch(b);
2632         }
2633
2634         prev_fanout = convert_num_notes_to_fanout(b->num_notes);
2635
2636         /* file_change* */
2637         while (command_buf.len > 0) {
2638                 if (skip_prefix(command_buf.buf, "M ", &v))
2639                         file_change_m(v, b);
2640                 else if (skip_prefix(command_buf.buf, "D ", &v))
2641                         file_change_d(v, b);
2642                 else if (skip_prefix(command_buf.buf, "R ", &v))
2643                         file_change_cr(v, b, 1);
2644                 else if (skip_prefix(command_buf.buf, "C ", &v))
2645                         file_change_cr(v, b, 0);
2646                 else if (skip_prefix(command_buf.buf, "N ", &v))
2647                         note_change_n(v, b, &prev_fanout);
2648                 else if (!strcmp("deleteall", command_buf.buf))
2649                         file_change_deleteall(b);
2650                 else if (skip_prefix(command_buf.buf, "ls ", &v))
2651                         parse_ls(v, b);
2652                 else if (skip_prefix(command_buf.buf, "cat-blob ", &v))
2653                         parse_cat_blob(v);
2654                 else {
2655                         unread_command_buf = 1;
2656                         break;
2657                 }
2658                 if (read_next_command() == EOF)
2659                         break;
2660         }
2661
2662         new_fanout = convert_num_notes_to_fanout(b->num_notes);
2663         if (new_fanout != prev_fanout)
2664                 b->num_notes = change_note_fanout(&b->branch_tree, new_fanout);
2665
2666         /* build the tree and the commit */
2667         store_tree(&b->branch_tree);
2668         oidcpy(&b->branch_tree.versions[0].oid,
2669                &b->branch_tree.versions[1].oid);
2670
2671         strbuf_reset(&new_data);
2672         strbuf_addf(&new_data, "tree %s\n",
2673                 oid_to_hex(&b->branch_tree.versions[1].oid));
2674         if (!is_null_oid(&b->oid))
2675                 strbuf_addf(&new_data, "parent %s\n",
2676                             oid_to_hex(&b->oid));
2677         while (merge_list) {
2678                 struct hash_list *next = merge_list->next;
2679                 strbuf_addf(&new_data, "parent %s\n",
2680                             oid_to_hex(&merge_list->oid));
2681                 free(merge_list);
2682                 merge_list = next;
2683         }
2684         strbuf_addf(&new_data,
2685                 "author %s\n"
2686                 "committer %s\n",
2687                 author ? author : committer, committer);
2688         if (encoding)
2689                 strbuf_addf(&new_data,
2690                         "encoding %s\n",
2691                         encoding);
2692         strbuf_addch(&new_data, '\n');
2693         strbuf_addbuf(&new_data, &msg);
2694         free(author);
2695         free(committer);
2696
2697         if (!store_object(OBJ_COMMIT, &new_data, NULL, &b->oid, next_mark))
2698                 b->pack_id = pack_id;
2699         b->last_commit = object_count_by_type[OBJ_COMMIT];
2700 }
2701
2702 static void parse_new_tag(const char *arg)
2703 {
2704         static struct strbuf msg = STRBUF_INIT;
2705         const char *from;
2706         char *tagger;
2707         struct branch *s;
2708         struct tag *t;
2709         uintmax_t from_mark = 0;
2710         struct object_id oid;
2711         enum object_type type;
2712         const char *v;
2713
2714         t = mem_pool_alloc(&fi_mem_pool, sizeof(struct tag));
2715         memset(t, 0, sizeof(struct tag));
2716         t->name = pool_strdup(arg);
2717         if (last_tag)
2718                 last_tag->next_tag = t;
2719         else
2720                 first_tag = t;
2721         last_tag = t;
2722         read_next_command();
2723
2724         /* from ... */
2725         if (!skip_prefix(command_buf.buf, "from ", &from))
2726                 die("Expected from command, got %s", command_buf.buf);
2727         s = lookup_branch(from);
2728         if (s) {
2729                 if (is_null_oid(&s->oid))
2730                         die("Can't tag an empty branch.");
2731                 oidcpy(&oid, &s->oid);
2732                 type = OBJ_COMMIT;
2733         } else if (*from == ':') {
2734                 struct object_entry *oe;
2735                 from_mark = parse_mark_ref_eol(from);
2736                 oe = find_mark(from_mark);
2737                 type = oe->type;
2738                 oidcpy(&oid, &oe->idx.oid);
2739         } else if (!get_oid(from, &oid)) {
2740                 struct object_entry *oe = find_object(&oid);
2741                 if (!oe) {
2742                         type = oid_object_info(the_repository, &oid, NULL);
2743                         if (type < 0)
2744                                 die("Not a valid object: %s", from);
2745                 } else
2746                         type = oe->type;
2747         } else
2748                 die("Invalid ref name or SHA1 expression: %s", from);
2749         read_next_command();
2750
2751         /* original-oid ... */
2752         parse_original_identifier();
2753
2754         /* tagger ... */
2755         if (skip_prefix(command_buf.buf, "tagger ", &v)) {
2756                 tagger = parse_ident(v);
2757                 read_next_command();
2758         } else
2759                 tagger = NULL;
2760
2761         /* tag payload/message */
2762         parse_data(&msg, 0, NULL);
2763
2764         /* build the tag object */
2765         strbuf_reset(&new_data);
2766
2767         strbuf_addf(&new_data,
2768                     "object %s\n"
2769                     "type %s\n"
2770                     "tag %s\n",
2771                     oid_to_hex(&oid), type_name(type), t->name);
2772         if (tagger)
2773                 strbuf_addf(&new_data,
2774                             "tagger %s\n", tagger);
2775         strbuf_addch(&new_data, '\n');
2776         strbuf_addbuf(&new_data, &msg);
2777         free(tagger);
2778
2779         if (store_object(OBJ_TAG, &new_data, NULL, &t->oid, 0))
2780                 t->pack_id = MAX_PACK_ID;
2781         else
2782                 t->pack_id = pack_id;
2783 }
2784
2785 static void parse_reset_branch(const char *arg)
2786 {
2787         struct branch *b;
2788
2789         b = lookup_branch(arg);
2790         if (b) {
2791                 oidclr(&b->oid);
2792                 oidclr(&b->branch_tree.versions[0].oid);
2793                 oidclr(&b->branch_tree.versions[1].oid);
2794                 if (b->branch_tree.tree) {
2795                         release_tree_content_recursive(b->branch_tree.tree);
2796                         b->branch_tree.tree = NULL;
2797                 }
2798         }
2799         else
2800                 b = new_branch(arg);
2801         read_next_command();
2802         parse_from(b);
2803         if (command_buf.len > 0)
2804                 unread_command_buf = 1;
2805 }
2806
2807 static void cat_blob_write(const char *buf, unsigned long size)
2808 {
2809         if (write_in_full(cat_blob_fd, buf, size) < 0)
2810                 die_errno("Write to frontend failed");
2811 }
2812
2813 static void cat_blob(struct object_entry *oe, struct object_id *oid)
2814 {
2815         struct strbuf line = STRBUF_INIT;
2816         unsigned long size;
2817         enum object_type type = 0;
2818         char *buf;
2819
2820         if (!oe || oe->pack_id == MAX_PACK_ID) {
2821                 buf = read_object_file(oid, &type, &size);
2822         } else {
2823                 type = oe->type;
2824                 buf = gfi_unpack_entry(oe, &size);
2825         }
2826
2827         /*
2828          * Output based on batch_one_object() from cat-file.c.
2829          */
2830         if (type <= 0) {
2831                 strbuf_reset(&line);
2832                 strbuf_addf(&line, "%s missing\n", oid_to_hex(oid));
2833                 cat_blob_write(line.buf, line.len);
2834                 strbuf_release(&line);
2835                 free(buf);
2836                 return;
2837         }
2838         if (!buf)
2839                 die("Can't read object %s", oid_to_hex(oid));
2840         if (type != OBJ_BLOB)
2841                 die("Object %s is a %s but a blob was expected.",
2842                     oid_to_hex(oid), type_name(type));
2843         strbuf_reset(&line);
2844         strbuf_addf(&line, "%s %s %"PRIuMAX"\n", oid_to_hex(oid),
2845                     type_name(type), (uintmax_t)size);
2846         cat_blob_write(line.buf, line.len);
2847         strbuf_release(&line);
2848         cat_blob_write(buf, size);
2849         cat_blob_write("\n", 1);
2850         if (oe && oe->pack_id == pack_id) {
2851                 last_blob.offset = oe->idx.offset;
2852                 strbuf_attach(&last_blob.data, buf, size, size);
2853                 last_blob.depth = oe->depth;
2854         } else
2855                 free(buf);
2856 }
2857
2858 static void parse_get_mark(const char *p)
2859 {
2860         struct object_entry *oe;
2861         char output[GIT_MAX_HEXSZ + 2];
2862
2863         /* get-mark SP <object> LF */
2864         if (*p != ':')
2865                 die("Not a mark: %s", p);
2866
2867         oe = find_mark(parse_mark_ref_eol(p));
2868         if (!oe)
2869                 die("Unknown mark: %s", command_buf.buf);
2870
2871         xsnprintf(output, sizeof(output), "%s\n", oid_to_hex(&oe->idx.oid));
2872         cat_blob_write(output, the_hash_algo->hexsz + 1);
2873 }
2874
2875 static void parse_cat_blob(const char *p)
2876 {
2877         struct object_entry *oe;
2878         struct object_id oid;
2879
2880         /* cat-blob SP <object> LF */
2881         if (*p == ':') {
2882                 oe = find_mark(parse_mark_ref_eol(p));
2883                 if (!oe)
2884                         die("Unknown mark: %s", command_buf.buf);
2885                 oidcpy(&oid, &oe->idx.oid);
2886         } else {
2887                 if (parse_oid_hex(p, &oid, &p))
2888                         die("Invalid dataref: %s", command_buf.buf);
2889                 if (*p)
2890                         die("Garbage after SHA1: %s", command_buf.buf);
2891                 oe = find_object(&oid);
2892         }
2893
2894         cat_blob(oe, &oid);
2895 }
2896
2897 static struct object_entry *dereference(struct object_entry *oe,
2898                                         struct object_id *oid)
2899 {
2900         unsigned long size;
2901         char *buf = NULL;
2902         const unsigned hexsz = the_hash_algo->hexsz;
2903
2904         if (!oe) {
2905                 enum object_type type = oid_object_info(the_repository, oid,
2906                                                         NULL);
2907                 if (type < 0)
2908                         die("object not found: %s", oid_to_hex(oid));
2909                 /* cache it! */
2910                 oe = insert_object(oid);
2911                 oe->type = type;
2912                 oe->pack_id = MAX_PACK_ID;
2913                 oe->idx.offset = 1;
2914         }
2915         switch (oe->type) {
2916         case OBJ_TREE:  /* easy case. */
2917                 return oe;
2918         case OBJ_COMMIT:
2919         case OBJ_TAG:
2920                 break;
2921         default:
2922                 die("Not a tree-ish: %s", command_buf.buf);
2923         }
2924
2925         if (oe->pack_id != MAX_PACK_ID) {       /* in a pack being written */
2926                 buf = gfi_unpack_entry(oe, &size);
2927         } else {
2928                 enum object_type unused;
2929                 buf = read_object_file(oid, &unused, &size);
2930         }
2931         if (!buf)
2932                 die("Can't load object %s", oid_to_hex(oid));
2933
2934         /* Peel one layer. */
2935         switch (oe->type) {
2936         case OBJ_TAG:
2937                 if (size < hexsz + strlen("object ") ||
2938                     get_oid_hex(buf + strlen("object "), oid))
2939                         die("Invalid SHA1 in tag: %s", command_buf.buf);
2940                 break;
2941         case OBJ_COMMIT:
2942                 if (size < hexsz + strlen("tree ") ||
2943                     get_oid_hex(buf + strlen("tree "), oid))
2944                         die("Invalid SHA1 in commit: %s", command_buf.buf);
2945         }
2946
2947         free(buf);
2948         return find_object(oid);
2949 }
2950
2951 static struct object_entry *parse_treeish_dataref(const char **p)
2952 {
2953         struct object_id oid;
2954         struct object_entry *e;
2955
2956         if (**p == ':') {       /* <mark> */
2957                 e = find_mark(parse_mark_ref_space(p));
2958                 if (!e)
2959                         die("Unknown mark: %s", command_buf.buf);
2960                 oidcpy(&oid, &e->idx.oid);
2961         } else {        /* <sha1> */
2962                 if (parse_oid_hex(*p, &oid, p))
2963                         die("Invalid dataref: %s", command_buf.buf);
2964                 e = find_object(&oid);
2965                 if (*(*p)++ != ' ')
2966                         die("Missing space after tree-ish: %s", command_buf.buf);
2967         }
2968
2969         while (!e || e->type != OBJ_TREE)
2970                 e = dereference(e, &oid);
2971         return e;
2972 }
2973
2974 static void print_ls(int mode, const unsigned char *hash, const char *path)
2975 {
2976         static struct strbuf line = STRBUF_INIT;
2977
2978         /* See show_tree(). */
2979         const char *type =
2980                 S_ISGITLINK(mode) ? commit_type :
2981                 S_ISDIR(mode) ? tree_type :
2982                 blob_type;
2983
2984         if (!mode) {
2985                 /* missing SP path LF */
2986                 strbuf_reset(&line);
2987                 strbuf_addstr(&line, "missing ");
2988                 quote_c_style(path, &line, NULL, 0);
2989                 strbuf_addch(&line, '\n');
2990         } else {
2991                 /* mode SP type SP object_name TAB path LF */
2992                 strbuf_reset(&line);
2993                 strbuf_addf(&line, "%06o %s %s\t",
2994                                 mode & ~NO_DELTA, type, hash_to_hex(hash));
2995                 quote_c_style(path, &line, NULL, 0);
2996                 strbuf_addch(&line, '\n');
2997         }
2998         cat_blob_write(line.buf, line.len);
2999 }
3000
3001 static void parse_ls(const char *p, struct branch *b)
3002 {
3003         struct tree_entry *root = NULL;
3004         struct tree_entry leaf = {NULL};
3005
3006         /* ls SP (<tree-ish> SP)? <path> */
3007         if (*p == '"') {
3008                 if (!b)
3009                         die("Not in a commit: %s", command_buf.buf);
3010                 root = &b->branch_tree;
3011         } else {
3012                 struct object_entry *e = parse_treeish_dataref(&p);
3013                 root = new_tree_entry();
3014                 oidcpy(&root->versions[1].oid, &e->idx.oid);
3015                 if (!is_null_oid(&root->versions[1].oid))
3016                         root->versions[1].mode = S_IFDIR;
3017                 load_tree(root);
3018         }
3019         if (*p == '"') {
3020                 static struct strbuf uq = STRBUF_INIT;
3021                 const char *endp;
3022                 strbuf_reset(&uq);
3023                 if (unquote_c_style(&uq, p, &endp))
3024                         die("Invalid path: %s", command_buf.buf);
3025                 if (*endp)
3026                         die("Garbage after path in: %s", command_buf.buf);
3027                 p = uq.buf;
3028         }
3029         tree_content_get(root, p, &leaf, 1);
3030         /*
3031          * A directory in preparation would have a sha1 of zero
3032          * until it is saved.  Save, for simplicity.
3033          */
3034         if (S_ISDIR(leaf.versions[1].mode))
3035                 store_tree(&leaf);
3036
3037         print_ls(leaf.versions[1].mode, leaf.versions[1].oid.hash, p);
3038         if (leaf.tree)
3039                 release_tree_content_recursive(leaf.tree);
3040         if (!b || root != &b->branch_tree)
3041                 release_tree_entry(root);
3042 }
3043
3044 static void checkpoint(void)
3045 {
3046         checkpoint_requested = 0;
3047         if (object_count) {
3048                 cycle_packfile();
3049         }
3050         dump_branches();
3051         dump_tags();
3052         dump_marks();
3053 }
3054
3055 static void parse_checkpoint(void)
3056 {
3057         checkpoint_requested = 1;
3058         skip_optional_lf();
3059 }
3060
3061 static void parse_progress(void)
3062 {
3063         fwrite(command_buf.buf, 1, command_buf.len, stdout);
3064         fputc('\n', stdout);
3065         fflush(stdout);
3066         skip_optional_lf();
3067 }
3068
3069 static char* make_fast_import_path(const char *path)
3070 {
3071         if (!relative_marks_paths || is_absolute_path(path))
3072                 return xstrdup(path);
3073         return git_pathdup("info/fast-import/%s", path);
3074 }
3075
3076 static void option_import_marks(const char *marks,
3077                                         int from_stream, int ignore_missing)
3078 {
3079         if (import_marks_file) {
3080                 if (from_stream)
3081                         die("Only one import-marks command allowed per stream");
3082
3083                 /* read previous mark file */
3084                 if(!import_marks_file_from_stream)
3085                         read_marks();
3086         }
3087
3088         import_marks_file = make_fast_import_path(marks);
3089         import_marks_file_from_stream = from_stream;
3090         import_marks_file_ignore_missing = ignore_missing;
3091 }
3092
3093 static void option_date_format(const char *fmt)
3094 {
3095         if (!strcmp(fmt, "raw"))
3096                 whenspec = WHENSPEC_RAW;
3097         else if (!strcmp(fmt, "rfc2822"))
3098                 whenspec = WHENSPEC_RFC2822;
3099         else if (!strcmp(fmt, "now"))
3100                 whenspec = WHENSPEC_NOW;
3101         else
3102                 die("unknown --date-format argument %s", fmt);
3103 }
3104
3105 static unsigned long ulong_arg(const char *option, const char *arg)
3106 {
3107         char *endptr;
3108         unsigned long rv = strtoul(arg, &endptr, 0);
3109         if (strchr(arg, '-') || endptr == arg || *endptr)
3110                 die("%s: argument must be a non-negative integer", option);
3111         return rv;
3112 }
3113
3114 static void option_depth(const char *depth)
3115 {
3116         max_depth = ulong_arg("--depth", depth);
3117         if (max_depth > MAX_DEPTH)
3118                 die("--depth cannot exceed %u", MAX_DEPTH);
3119 }
3120
3121 static void option_active_branches(const char *branches)
3122 {
3123         max_active_branches = ulong_arg("--active-branches", branches);
3124 }
3125
3126 static void option_export_marks(const char *marks)
3127 {
3128         export_marks_file = make_fast_import_path(marks);
3129 }
3130
3131 static void option_cat_blob_fd(const char *fd)
3132 {
3133         unsigned long n = ulong_arg("--cat-blob-fd", fd);
3134         if (n > (unsigned long) INT_MAX)
3135                 die("--cat-blob-fd cannot exceed %d", INT_MAX);
3136         cat_blob_fd = (int) n;
3137 }
3138
3139 static void option_export_pack_edges(const char *edges)
3140 {
3141         if (pack_edges)
3142                 fclose(pack_edges);
3143         pack_edges = xfopen(edges, "a");
3144 }
3145
3146 static int parse_one_option(const char *option)
3147 {
3148         if (skip_prefix(option, "max-pack-size=", &option)) {
3149                 unsigned long v;
3150                 if (!git_parse_ulong(option, &v))
3151                         return 0;
3152                 if (v < 8192) {
3153                         warning("max-pack-size is now in bytes, assuming --max-pack-size=%lum", v);
3154                         v *= 1024 * 1024;
3155                 } else if (v < 1024 * 1024) {
3156                         warning("minimum max-pack-size is 1 MiB");
3157                         v = 1024 * 1024;
3158                 }
3159                 max_packsize = v;
3160         } else if (skip_prefix(option, "big-file-threshold=", &option)) {
3161                 unsigned long v;
3162                 if (!git_parse_ulong(option, &v))
3163                         return 0;
3164                 big_file_threshold = v;
3165         } else if (skip_prefix(option, "depth=", &option)) {
3166                 option_depth(option);
3167         } else if (skip_prefix(option, "active-branches=", &option)) {
3168                 option_active_branches(option);
3169         } else if (skip_prefix(option, "export-pack-edges=", &option)) {
3170                 option_export_pack_edges(option);
3171         } else if (!strcmp(option, "quiet")) {
3172                 show_stats = 0;
3173         } else if (!strcmp(option, "stats")) {
3174                 show_stats = 1;
3175         } else if (!strcmp(option, "allow-unsafe-features")) {
3176                 ; /* already handled during early option parsing */
3177         } else {
3178                 return 0;
3179         }
3180
3181         return 1;
3182 }
3183
3184 static void check_unsafe_feature(const char *feature, int from_stream)
3185 {
3186         if (from_stream && !allow_unsafe_features)
3187                 die(_("feature '%s' forbidden in input without --allow-unsafe-features"),
3188                     feature);
3189 }
3190
3191 static int parse_one_feature(const char *feature, int from_stream)
3192 {
3193         const char *arg;
3194
3195         if (skip_prefix(feature, "date-format=", &arg)) {
3196                 option_date_format(arg);
3197         } else if (skip_prefix(feature, "import-marks=", &arg)) {
3198                 check_unsafe_feature("import-marks", from_stream);
3199                 option_import_marks(arg, from_stream, 0);
3200         } else if (skip_prefix(feature, "import-marks-if-exists=", &arg)) {
3201                 check_unsafe_feature("import-marks-if-exists", from_stream);
3202                 option_import_marks(arg, from_stream, 1);
3203         } else if (skip_prefix(feature, "export-marks=", &arg)) {
3204                 check_unsafe_feature(feature, from_stream);
3205                 option_export_marks(arg);
3206         } else if (!strcmp(feature, "get-mark")) {
3207                 ; /* Don't die - this feature is supported */
3208         } else if (!strcmp(feature, "cat-blob")) {
3209                 ; /* Don't die - this feature is supported */
3210         } else if (!strcmp(feature, "relative-marks")) {
3211                 relative_marks_paths = 1;
3212         } else if (!strcmp(feature, "no-relative-marks")) {
3213                 relative_marks_paths = 0;
3214         } else if (!strcmp(feature, "done")) {
3215                 require_explicit_termination = 1;
3216         } else if (!strcmp(feature, "force")) {
3217                 force_update = 1;
3218         } else if (!strcmp(feature, "notes") || !strcmp(feature, "ls")) {
3219                 ; /* do nothing; we have the feature */
3220         } else {
3221                 return 0;
3222         }
3223
3224         return 1;
3225 }
3226
3227 static void parse_feature(const char *feature)
3228 {
3229         if (seen_data_command)
3230                 die("Got feature command '%s' after data command", feature);
3231
3232         if (parse_one_feature(feature, 1))
3233                 return;
3234
3235         die("This version of fast-import does not support feature %s.", feature);
3236 }
3237
3238 static void parse_option(const char *option)
3239 {
3240         if (seen_data_command)
3241                 die("Got option command '%s' after data command", option);
3242
3243         if (parse_one_option(option))
3244                 return;
3245
3246         die("This version of fast-import does not support option: %s", option);
3247 }
3248
3249 static void git_pack_config(void)
3250 {
3251         int indexversion_value;
3252         int limit;
3253         unsigned long packsizelimit_value;
3254
3255         if (!git_config_get_ulong("pack.depth", &max_depth)) {
3256                 if (max_depth > MAX_DEPTH)
3257                         max_depth = MAX_DEPTH;
3258         }
3259         if (!git_config_get_int("pack.indexversion", &indexversion_value)) {
3260                 pack_idx_opts.version = indexversion_value;
3261                 if (pack_idx_opts.version > 2)
3262                         git_die_config("pack.indexversion",
3263                                         "bad pack.indexversion=%"PRIu32, pack_idx_opts.version);
3264         }
3265         if (!git_config_get_ulong("pack.packsizelimit", &packsizelimit_value))
3266                 max_packsize = packsizelimit_value;
3267
3268         if (!git_config_get_int("fastimport.unpacklimit", &limit))
3269                 unpack_limit = limit;
3270         else if (!git_config_get_int("transfer.unpacklimit", &limit))
3271                 unpack_limit = limit;
3272
3273         git_config(git_default_config, NULL);
3274 }
3275
3276 static const char fast_import_usage[] =
3277 "git fast-import [--date-format=<f>] [--max-pack-size=<n>] [--big-file-threshold=<n>] [--depth=<n>] [--active-branches=<n>] [--export-marks=<marks.file>]";
3278
3279 static void parse_argv(void)
3280 {
3281         unsigned int i;
3282
3283         for (i = 1; i < global_argc; i++) {
3284                 const char *a = global_argv[i];
3285
3286                 if (*a != '-' || !strcmp(a, "--"))
3287                         break;
3288
3289                 if (!skip_prefix(a, "--", &a))
3290                         die("unknown option %s", a);
3291
3292                 if (parse_one_option(a))
3293                         continue;
3294
3295                 if (parse_one_feature(a, 0))
3296                         continue;
3297
3298                 if (skip_prefix(a, "cat-blob-fd=", &a)) {
3299                         option_cat_blob_fd(a);
3300                         continue;
3301                 }
3302
3303                 die("unknown option --%s", a);
3304         }
3305         if (i != global_argc)
3306                 usage(fast_import_usage);
3307
3308         seen_data_command = 1;
3309         if (import_marks_file)
3310                 read_marks();
3311 }
3312
3313 int cmd_main(int argc, const char **argv)
3314 {
3315         unsigned int i;
3316
3317         if (argc == 2 && !strcmp(argv[1], "-h"))
3318                 usage(fast_import_usage);
3319
3320         setup_git_directory();
3321         reset_pack_idx_option(&pack_idx_opts);
3322         git_pack_config();
3323
3324         alloc_objects(object_entry_alloc);
3325         strbuf_init(&command_buf, 0);
3326         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
3327         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
3328         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
3329         marks = mem_pool_calloc(&fi_mem_pool, 1, sizeof(struct mark_set));
3330
3331         /*
3332          * We don't parse most options until after we've seen the set of
3333          * "feature" lines at the start of the stream (which allows the command
3334          * line to override stream data). But we must do an early parse of any
3335          * command-line options that impact how we interpret the feature lines.
3336          */
3337         for (i = 1; i < argc; i++) {
3338                 const char *arg = argv[i];
3339                 if (*arg != '-' || !strcmp(arg, "--"))
3340                         break;
3341                 if (!strcmp(arg, "--allow-unsafe-features"))
3342                         allow_unsafe_features = 1;
3343         }
3344
3345         global_argc = argc;
3346         global_argv = argv;
3347
3348         rc_free = mem_pool_alloc(&fi_mem_pool, cmd_save * sizeof(*rc_free));
3349         for (i = 0; i < (cmd_save - 1); i++)
3350                 rc_free[i].next = &rc_free[i + 1];
3351         rc_free[cmd_save - 1].next = NULL;
3352
3353         start_packfile();
3354         set_die_routine(die_nicely);
3355         set_checkpoint_signal();
3356         while (read_next_command() != EOF) {
3357                 const char *v;
3358                 if (!strcmp("blob", command_buf.buf))
3359                         parse_new_blob();
3360                 else if (skip_prefix(command_buf.buf, "commit ", &v))
3361                         parse_new_commit(v);
3362                 else if (skip_prefix(command_buf.buf, "tag ", &v))
3363                         parse_new_tag(v);
3364                 else if (skip_prefix(command_buf.buf, "reset ", &v))
3365                         parse_reset_branch(v);
3366                 else if (skip_prefix(command_buf.buf, "ls ", &v))
3367                         parse_ls(v, NULL);
3368                 else if (skip_prefix(command_buf.buf, "cat-blob ", &v))
3369                         parse_cat_blob(v);
3370                 else if (skip_prefix(command_buf.buf, "get-mark ", &v))
3371                         parse_get_mark(v);
3372                 else if (!strcmp("checkpoint", command_buf.buf))
3373                         parse_checkpoint();
3374                 else if (!strcmp("done", command_buf.buf))
3375                         break;
3376                 else if (starts_with(command_buf.buf, "progress "))
3377                         parse_progress();
3378                 else if (skip_prefix(command_buf.buf, "feature ", &v))
3379                         parse_feature(v);
3380                 else if (skip_prefix(command_buf.buf, "option git ", &v))
3381                         parse_option(v);
3382                 else if (starts_with(command_buf.buf, "option "))
3383                         /* ignore non-git options*/;
3384                 else
3385                         die("Unsupported command: %s", command_buf.buf);
3386
3387                 if (checkpoint_requested)
3388                         checkpoint();
3389         }
3390
3391         /* argv hasn't been parsed yet, do so */
3392         if (!seen_data_command)
3393                 parse_argv();
3394
3395         if (require_explicit_termination && feof(stdin))
3396                 die("stream ends early");
3397
3398         end_packfile();
3399
3400         dump_branches();
3401         dump_tags();
3402         unkeep_all_packs();
3403         dump_marks();
3404
3405         if (pack_edges)
3406                 fclose(pack_edges);
3407
3408         if (show_stats) {
3409                 uintmax_t total_count = 0, duplicate_count = 0;
3410                 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
3411                         total_count += object_count_by_type[i];
3412                 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
3413                         duplicate_count += duplicate_count_by_type[i];
3414
3415                 fprintf(stderr, "%s statistics:\n", argv[0]);
3416                 fprintf(stderr, "---------------------------------------------------------------------\n");
3417                 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
3418                 fprintf(stderr, "Total objects:   %10" PRIuMAX " (%10" PRIuMAX " duplicates                  )\n", total_count, duplicate_count);
3419                 fprintf(stderr, "      blobs  :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB], delta_count_attempts_by_type[OBJ_BLOB]);
3420                 fprintf(stderr, "      trees  :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE], delta_count_attempts_by_type[OBJ_TREE]);
3421                 fprintf(stderr, "      commits:   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT], delta_count_attempts_by_type[OBJ_COMMIT]);
3422                 fprintf(stderr, "      tags   :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG], delta_count_attempts_by_type[OBJ_TAG]);
3423                 fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
3424                 fprintf(stderr, "      marks:     %10" PRIuMAX " (%10" PRIuMAX " unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
3425                 fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
3426                 fprintf(stderr, "Memory total:    %10" PRIuMAX " KiB\n", (tree_entry_allocd + fi_mem_pool.pool_alloc + alloc_count*sizeof(struct object_entry))/1024);
3427                 fprintf(stderr, "       pools:    %10lu KiB\n", (unsigned long)((tree_entry_allocd + fi_mem_pool.pool_alloc) /1024));
3428                 fprintf(stderr, "     objects:    %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
3429                 fprintf(stderr, "---------------------------------------------------------------------\n");
3430                 pack_report();
3431                 fprintf(stderr, "---------------------------------------------------------------------\n");
3432                 fprintf(stderr, "\n");
3433         }
3434
3435         return failure ? 1 : 0;
3436 }